nature methods | VOL.7 NO.10 | OCTOBER 2010 | 825
Unconditioned exploration models of fear and anxiety assess
caution or avoidance in potentially threatening situations2,9; urina-
tion and defecation are measures of emotionality associated with
anxiety or stress10. To assess anxiety-related behavior in antici-
pation of handling (similar to the use of human threat to assess
anxiety in nonhuman primates11), we assessed voluntary approach
and interaction with the handler immediately before and after daily
handling using the same handling method, as well as urination and
defecation during the handling procedure itself. We also compared
anxiety in an elevated plus maze when mice were familiar with
one of the three handling methods (Online Methods); in this test,
anxiety is evident from reduced entry into two open arms without
walls to provide physical protection, and an increased frequency of
protected stretched attend postures in which the mouse stretched
forward into an open arm but then retracted back into the pro-
tected center or closed arms of the maze.
As strains and sexes can differ substantially in anxiety and
stress responsiveness12,13, we assessed responses of both sexes
for two common inbred mouse strains, BALB/c and C57BL/6,
and the outbred strain ICR(CD-1). Mice of all strains developed
a consistent response to the three handling methods over nine
daily 60-s handling sessions (Fig. 1 and Supplementary Tables 1
and 2). Those handled by the tail (mouse body weight was sup-
ported on the hand or arm) showed the lowest voluntary inter-
action with the handler. By contrast, voluntary interaction was
prolonged among all tunnel-handled mice and among cupped
ICR(CD-1) mice and cupped BALB/c females; other cupped
mice showed intermediate responses. Prolonged interaction
reflected the willingness of many tunnel- or cup-handled mice
to voluntarily enter tunnels or climb on the hand and arm of
taming anxiety in
Jane L Hurst & Rebecca S West
routine laboratory animal handling has profound effects on
their anxiety and stress responses, but little is known about
the impact of handling method. We found that picking up mice
by the tail induced aversion and high anxiety, whereas use of
tunnels or open hand led to voluntary approach, low anxiety and
acceptance of physical restraint. using the latter methods, one
can minimize a widespread source of anxiety in laboratory mice.
Routine handling of laboratory animals is an essential but frequently
ignored component of animal experiments that has considerable
potential to influence anxiety and aversion to human approach and
contact. Laboratory mice seek to avoid capture and restraint unless
they have learned that handling is not harmful. Prior handling expe-
rience (which includes routine maintenance) can have positive or
negative effects on stress responses that influence experiments,
depending on the animal’s experience during handling1–5. We show
here that handling method itself is critical as it can induce fear and
anxiety responses to human contact.
The most common method used to capture and handle laboratory
mice is to pick up and restrain the mouse by its tail, a method usually
specified in standardized protocols6–8. We compared this with two alter-
natives (Supplementary Movies 1–3). Mice handled by ‘tunnel’ walked
into a clear acrylic home cage tunnel brought toward them (present in
all cages regardless of treatment) and were lifted without direct contact.
Mice that handlers ‘cupped’ were scooped up and allowed to walk freely
over the handler’s open gloved hands without direct physical restraint;
as mice unfamiliar with this technique immediately jump away, han-
dlers closed their hands loosely around the mouse for up to 30 s on first
experience, until the mouse’s attempts to escape declined. Subsequently
the handlers could pick up the mouse on open hands.
Mammalian Behaviour and Evolution Group, Institute of Integrative Biology, University of Liverpool, Leahurst Campus, Neston, UK. Correspondence should be
addressed to J.L.H. (email@example.com).
Received 30 June; accepted 9 august; published online 12 septembeR 2010; doi:10.1038/nmeth.1500
F2,18 = 39.9, P < 0.0001
P < 0.0001
2 = 20.5
P < 0.0001
2 = 18.6
P = 0.002
2 = 12.2
F2,18 = 47.8, P < 0.0001F2,18 = 20.0, P < 0.0001
Tail Tunnel Cup
during test time (%)
Tail Tunnel CupTail Tunnel Cup
Tail Tunnel Cup Tail Tunnel CupTail Tunnel Cup
figure 1 | Effect of handling method on voluntary interaction with the
handler and behavior in an elevated plus maze, tested during the dark
phase of the diurnal cycle. (a) Voluntary interaction in the ninth handling
session immediately after handling by the tail, home cage tunnel or
cupped on open hands among male and female BALB/c (left), C57BL/6
(middle) and ICR(CD-1) (right) mice. Error bars, s.e.m., n = 4 cages.
(b) Mean frequency of protected stretched attend postures during a 5-min
elevated plus maze test after nine handling sessions. Error bars, s.e.m.,
n = 8 mice. The overall effect of handling method is shown for parametric
(F) or nonparametric (χ2) ANOVAs; planned contrasts to tail method:
*P < 0.05, **P < 0.01, ***P < 0.005.
© 2010 Nature America, Inc. All rights reserved.
826 | VOL.7 NO.10 | OCTOBER 2010 | nature methods
the handler by session 9; this response developed quickly toward
familiar home cage tunnels and more slowly to cupping on the
open hand (Supplementary Fig. 1). This contrasted strongly
with brief and cautious approaches among mice handled by
the tail (Supplementary Movies 4–6). Compared to tunnel or
cup handling, tail handling also induced greater urination and
defecation during handling (Supplementary Fig. 2), together
with a higher frequency of protected stretched attend postures
(Fig. 1) and fewer open arm entries on the elevated plus maze
(Supplementary Fig. 3). Responses were very similar in the light
and dark phases of the diurnal cycle and were not influenced by
individual differences or prior experience of handlers, and any
interactions between sex and handling method were only minor
(Supplementary Figs. 1–5 and Supplementary Tables 1–3).
However, handling tunnels provide an alternative means to accus-
tom mice to being picked up. Mice previously experienced with
tunnel handling (16 sessions) responded as positively to their
first experience of being cupped on the hand (BALB/c mean ±
s.e.m., 42.5 ± 5.2% of test time interacting with the handler) as
those preconditioned to cupping (32.7 ± 6.0%; t-test: t12 = 1.24,
P = 0.24). Mice readily sat on the hand without jumping after
five sessions of tunnel handling (Supplementary Movie 7), pro-
viding a convenient combined handling method that reduced
anxiety in the elevated plus maze and in response to the handler
(Supplementary Fig. 4 and Supplementary Movie 8).
Restraining and lifting mice by the tail for abdominal inspec-
tion was not aversive if we first picked up the mice and placed
them on the hand by tunnel or cupping (Supplementary Fig. 6).
Aversion to tail handling thus was induced by being captured
and picked up by the tail, probably stemming from a naturally
selected ancestral imperative to avoid capture when fleeing from
predators. For many procedures, mice need to be restrained
more securely by the scruff of the neck. Scruff restraint did not
reverse the taming effects of being handled by tunnel or by cup-
ping as this did not stimulate increased avoidance of the handler
(Fig. 2 and Supplementary Table 4), suggesting that anxiety in
anticipation of handling was not increased, whereas tail-handled
mice continued to avoid handler interaction.
The notable difference in response induced by these alternative
routine handling methods has not previously been recognized to
our knowledge, maybe because picking up mice by the tail is so
widely used in laboratories that the aversive and anxious response is
perceived as ‘normal’. Consistent use of methods that do not induce
strong anxiety responses will minimize confounding responses
owing to routine handling before and during experiments, reducing
the need to standardize handling experience and timing. Conversely,
when a strong anxiety response is required for specific research pur-
poses, it may be important to ensure that mice are handled only by
the tail and are not tamed to human contact accidentally during
routine husbandry procedures. In addition to providing more robust
scientific outcomes, appropriate choice of handling method could
enhance the welfare of the many millions of mice that are housed
and handled in laboratories worldwide.
Methods and any associated references are available in the online
version of the paper at http://www.nature.com/naturemethods/.
Note: Supplementary information is available on the Nature Methods website.
We thank R. Beynon, F. Fair, J. Waters, S. Roberts, M. Garratt, J.-F. Lemaitre,
S. Ramm, M. Thom, S. Rounding and R. Humphries for practical assistance, and
R. Beynon, P. Stockley, S. Roberts and J. Rees for comments on a draft
manuscript. The study was funded by the Biotechnology and Biological Sciences
Research Council, the National Centre for the Replacement, Refinement and
Reduction of Animals in Research, the Wellcome Trust and the Association for
the Study of Animal Behaviour.
J.L.H. gained the funding and designed the project with contributions from
R.S.W.; J.L.H. and R.S.W. collected data; and J.L.H. supervised all aspects of the
work, analyzed the data and wrote the manuscript.
comPeting financial interests
The authors declare no competing financial interests.
Published online at http://www.nature.com/naturemethods/.
reprints and permissions information is available online at http://npg.nature.
1. Holson, R.R., Scallet, A.C., Ali, S.F. & Turner, B.B. Physiol. Behav. 49,
2. Rodgers, R.J. & Dalvi, A. Neurosci. Biobehav. Rev. 21, 801–810 (1997).
3. Schmitt, U. & Hiemke, C. Pharmacol. Biochem. Behav. 59, 807–811 (1998).
4. Balcombe, J.P., Barnard, N.D. & Sandusky, C. Contemp. Top. Lab. Anim.
Sci. 43, 42–51 (2004).
5. Leussis, M.P. & Heinrichs, S.C. Epilepsia 47, 801–804 (2006).
6. Wahlsten, D. et al. J. Neurobiol. 54, 283–311 (2003).
7. Deacon, R.M. Nat. Protoc. 1, 936–946 (2006).
8. Leach, M.C. & Main, D.C. Anim. Welf. 17, 171–187 (2008).
9. Ramos, A. Trends Pharmacol. Sci. 29, 493–498 (2008).
10. Henderson, N.D., Turri, M.G., DeFries, J.C. & Flint, J. Behav. Genet. 34,
11. Barros, M. & Tomaz, C. Neurosci. Biobehav. Rev. 26, 187–201 (2002).
12. Augustsson, H., Dahlborn, K. & Meyerson, B.J. Physiol. Behav. 84,
13. Roy, V., Merali, Z., Poulter, M.O. & Anisman, H. Behav. Brain Res. 185,
F2,18 = 20.4, P < 0.0001
Voluntary interaction during
test time (%)
F2,18 = 99.3, P < 0.0001
Voluntary interaction during
test time (%)
F2,18 = 26.8, P < 0.0001
Voluntary interaction during
test time (%)
F2,18 = 22.2, P < 0.0001
F2,18 = 68.6, P < 0.0001
F2,18 = 21.5, P < 0.0001
figure 2 | Voluntary interaction with the handler after restraint by the scruff for mice experienced with different handling methods. After 11 previous
handling sessions by tail, home cage tunnel or cupping on the open hand during the dark period, each mouse was picked up by its familiar method,
restrained by the scruff and held on its back in the palm of the hand for 10 s (session 12). Voluntary interaction was assessed before and after scruff
handling (session 12) and before and after normal handling the following day (session 13). Error bars, s.e.m., n = 8 cages (4 for each sex) for BALB/c
(left), C57BL/6 (middle) and ICR(CD-1) (right) mice. The overall effect of handling method is shown for repeated measure ANOVAs.
© 2010 Nature America, Inc. All rights reserved.
Subject animals. A total of five separate batches of mice were
tested. Batches 1–4 were obtained from an approved supplier
(Harlan; mice aged 3–4 weeks); batch 5 mice were bred in-house
from mice obtained from Harlan (parents handled by the standard
tail method). On arrival or at weaning, mice were housed in single
sex groups of two to six mice in 45 × 28 × 13 cm polycarbonate cages
(MB1; North Kent Plastics) on Corn Cob Absorb 10/14 bedding
with shredded paper nest material, ad libitum food (Lab Diet
5002 Certified Rodent Diet; Purina Mills) and water. Before the
experiment, mice were handled using the standard tail method
to transfer them between cages during cage cleaning. Mice were
marked for individual identification on the shoulders or rump
using hair dye (Clairol Nice ‘n’ Easy Natural Black for BALB/c
and ICR(CD-1), Clairol Nice ‘n’ Easy Brilliant Blond for C57BL/6;
Bristol-Myers Co. Ltd.) and split into single sex dyads in 43 ×
11.5 × 12 cm cages (M3, North Kent Plastics), each containing a
clear acrylic tunnel (50 mm diameter, 200 mm long) 3 d before
the start of testing (mice aged 8–10 weeks). Pilot tests confirmed
that the marking procedure did not influence the response to dif-
ferent handling methods. Batches 1 and 2 each consisted of eight
cages of two BALB/c females assigned randomly to each of three
handling methods (48 females per batch). To assess the generality
of response to the three handling methods across sexes and
strains, four cages of two mice per sex and strain were assigned
randomly to each of the three methods for outbred ICR(CD-1)
mice (batch 3, 48 mice) and inbred BALB/c or C57BL/6 mice
(batch 4, 48 mice per strain). Batches 1–4 were all handled and
tested during the dark phase under red lighting so that mice would
be naturally active during testing. Mice in batch 5 were handled
and tested during the light phase, when many laboratory animals
are normally handled, with three to four cages of two mice per sex
and strain assigned randomly to each of three (BALB/c, 42 mice)
or four handling methods (C57BL/6, 64 mice). In each batch
and strain, cages handled by each method were arranged in a
balanced design on the cage rack. Throughout, mice were main-
tained under a reversed 12-h light–12 h dark schedule (white lights
on 21:00–09:00) or a nonreversed light schedule (09:00–21:00,
batch 5 only). All procedures were noninvasive and involved
standard husbandry practices, so did not require any specific
licences (University of Liverpool Animal Welfare Committee).
Handling methods. From the start of testing, mice were handled
only by their designated method. Before handling, nest material
and the home cage tunnel were removed from the cage. For tail
handling, the base of the tail was grasped between thumb and
forefinger and the mouse gently lifted onto the opposite gloved
hand or laboratory coat sleeve and held there by the tail for 30 s
before release back into the cage. For tunnel handling, the mouse
was guided into the home cage tunnel (soiled with familiar scent)
as it was brought toward it, and the tunnel lifted above the cage
and held for 30 s. The handler’s hands were sometimes loosely
cupped over the tunnel ends to prevent escape on the first 1 or 2 d
before mice became accustomed to tunnel handling. A cupped
mouse was scooped up on one or both open hands and allowed
to sit or walk over the hands for 30 s without other physical
restraint. To prevent mice leaping from the hands, for the first
experience of cupping the hands were closed loosely around the
mouse for a maximum of 30 s until attempts to escape declined,
after which the hands were opened. This was not necessary after
the first handling session. All batches compared responses to
these three different methods. A fourth handling method, which
combined tunnel and cup handling, was tested using C57BL/6
mice (batch 5) to establish an alternative way to prevent mice
initially leaping from the hands during cup handling. Relatively
inexperienced handlers may, at first, find cup handling of
untamed mice quite difficult for some strain and age combina-
tions that are more likely to jump from the hands, or handlers
may be concerned about being bitten when confining untamed
mice between the hands (though no cupped mice attempted to
bite in this study). For the first five handling sessions, mice were
picked up in their home cage tunnel; on the sixth session they
were picked up in the tunnel then tipped backwards onto the
open hand for 30 s, after which they were picked up directly by
cupping on the open hand. Supplementary Movies 1–3 and 7
demonstrate each handling method.
In each handling session, the handler picked up each mouse
in turn by the designated method and held it for 30 s. After
moving away from the open cage for 60 s, the mice were han-
dled again so that each mouse was picked up twice and held for
a total of 60 s. Handlers wore laboratory coats contaminated
with mouse scent from previous handling sessions and clean
close-fitting nitrile surgical gloves that were rubbed in bedding
soiled by the same strain and sex of mice before the start of each
handling session. During the first nine handling sessions, the
occurrence of any urination or defecation each time a mouse
was picked up and held was recorded (maximum score per mea-
sure for each mouse = 18).
Handling and test schedule. The first batch of BALB/c females
was handled by a single handler (n = 8 cages per handling method)
for 16 separate handling sessions. To assess variability in response
across handlers, the second batch of BALB/c females was handled
by eight different handlers that varied widely in their prior expe-
rience of handling rodents (each individual handled three cages
of mice using one cage per handling method; cages were assigned
randomly in a balanced design). Two handlers had no prior expe-
rience, three handlers were researchers with 3–4 years of experi-
ence working with rodents, three handlers were full-time animal
care staff and a researcher with substantial rodent handling expe-
rience. Handlers were given a practice session using stock mice
before the start of the experiment after demonstration of each
method and test procedure. Response toward the handler was
assessed over nine daily handling sessions using all batches of
mice, with responses tested during handling sessions 1 (day 1),
5 (day 5) and 9 (day 11 or 12). Experienced batch 1 mice were
also used to test response after 11 d without handling, an open
field test of anxiety, the effect of mouse-soiled protective cloth-
ing on response toward the handler (data not shown) and the
response to cup handling among mice previously handled by
tail or tunnel. Experienced batch 2 mice were used to test anxi-
ety in an elevated plus maze, response to an unfamiliar handler
and response to tail manipulation for abdominal inspection by
the unfamiliar handler. Batch 3 outbred mice were handled by
one of two handlers (one experienced and one inexperienced;
data were combined as handler had no effects), and batches 4–5
were handled by a single experienced handler. Details of the
handling sessions and tests conducted on each batch of mice,
© 2010 Nature America, Inc. All rights reserved.
and their timing are summarized in Supplementary Table 5.
In each batch, order of testing cages assigned to the same
method was randomized, and the order was balanced across the
three different methods throughout. Gloves were changed
between cages housing mice of the opposite sex. Cage clean-
ing was carried out fortnightly at the end of a week to avoid
interference with testing, and mice were transferred to a
clean cage by their familiar handling method.
Response to the handler. To assess behavior in anticipation of
handling, we assessed voluntary approach and interaction with
the handler before and after handling on specified test days. After
removal of the cage lid, nest material and home cage tunnel, the
handler stood unmoving for 60 s directly in front of the cage.
A gloved hand (tail and cup methods) or gloved hand holding
the home cage tunnel (tunnel method) was then held resting on
the substrate in the front half of the cage without moving for an
additional 60 s to assess voluntary interaction. Both mice were
then handled by the designated method as described above, the
handler stood back from the cage for 60 s, and then repeated
the two 60-s tests. Trials were recorded on video or DVD for
transcription by a single observer using an event timing program
(blinded with respect to test on the second batch of mice). For
each test, we measured the time spent in the front half of the cage
by each mouse and time spent interacting at close contact with
the introduced hand or hand and tunnel (time spent sniffing the
handling device, paws on, climbing on, inside tunnel or chewing
the glove were summed for analysis).
Anxiety tests. Mice were tested using an elevated plus maze after
seven or nine handling sessions (Supplementary Table 5). Mice
were delivered by their familiar handler and handling method
to the center of the maze (arms 30 × 5 cm with side walls 15 cm
high on the two closed arms, elevated 57 cm above the ground),
facing an open arm, for a 5 min test. Mice handled by the com-
bined tunnel and cup method were delivered to the maze in their
home cage tunnel. We scored the number of entries and total
time when all four feet were in each arm of the maze (summed
for open or closed arms), the frequencies of protected stretched
attend postures into the open arms from the central hub or closed
arms, and unprotected stretched attends on the open arms2. At
the end of the test, mice were returned to their home cage using
their familiar method of handling and the arena cleaned with 70%
ethanol and dried with a paper towel. Mice from the same cage
were tested successively on the same day.
Tail manipulation for abdominal inspection. To assess response
to lifting the mouse by its tail to examine its ventral surface and
anogenital area, BALB/c females (batch 2) and outbred ICR(CD-1)
mice of both sexes (batch 3) were captured using their familiar
method (following 13 or 10 previous handling sessions, respec-
tively) and placed on the handler’s open gloved hand (tunnel mice
were tipped backwards out of the tunnel on to the hand). The tail
base was then grasped between the thumb and forefinger of the
other hand and the back legs of the mouse lifted off the hand to
expose the ventrum and held for 10 s. Response to the handler
was assessed immediately before and after handling.
Scruff handling. To assess response to the experience of being
restrained by the loose skin of the scruff, ICR(CD-1), BALB/c and
C57BL/6 mice in batches 3 and 4 were captured using their fami-
liar method after 11 previous handling sessions and placed on the
bars of a clean cage top. The loose skin of the scruff was grasped
between thumb and forefinger to immobilise the animal in the
hand, and the mouse held on its back above the home cage for 10 s
before being released back into the cage. Response to the handler
was assessed immediately before and after scruff handling, and
before and after handling by their normal method the following
day to assess how quickly mice recovered from scruff handling.
Response to cup handling. The final test of BALB/c batch 1
females assessed response to the familiar handler before and
after all mice were handled using the cupping method. This was
designed to test how mice previously handled indirectly using a
tunnel would respond to direct physical contact with the hand.
Response to the handler was assessed before and after cup han-
dling by inserting a gloved hand only into the cage for all mice,
tested on day 45 after 15 prior handling sessions with the familiar
Data analysis. For tests in which all cages were handled by a
single handler, data were averaged for the two mice in each cage;
repeated measures ANOVAs examined response immediately
before and after handling as a within-subjects effect and handling
method and sex as between-subjects effects. To assess response
to the handler using nine different handlers, data were averaged
across all mice handled by each method for each handler (that
is, two mice per cage for eight handlers in batch 2, two mice
times eight cages for one handler in batch 1); repeated measures
ANOVAs examined handling method and response immediately
before and after handling as within-subject effects per handler.
For elevated plus maze tests where mice were removed from their
home cage and tested individually, the effect of handling method
was examined using parametric or nonparametric14 ANOVAs,
with sex as an additional factor. In these tests, data from each
mouse were treated as independent to follow normal convention.
However, confirmatory repeated measures ANOVAs of mean
response per cage found similar statistical significance between
handling methods (for example, for the data shown in Figure 1
and Supplementary Figure 3, mean response per cage for pro-
tected stretch attend: BALB/c, χ2 = 14.2, P = 0.001; C57BL/6,
χ2 = 12.3, P = 0.002; ICR, χ2 = 10.5, P = 0.005; open arm entries:
BALB/c, F2,18 = 0.29, P = 0.76; C57BL/6, F2,18 = 6.19, P = 0.009;
ICR, χ2 = 7.0, P = 0.030). Within each ANOVA, planned contrasts
tested whether the response to each alternative handling method
differed significantly from the standard tail handling method
(Mann-Whitney U tests were used for nonparametric analyses).
Nonparametric tests14 were used when graphical inspection of
data and Kolmogorov-Smirnov and Shapiro-Wilk tests indicated
that data did not approximate normality and data could not be
transformed to meet assumptions of parametric analyses.
14. Meddis, R. Statistics Using Ranks: A Unified Approach. (Basil Blackwell
Publisher Ltd, Oxford, 1984.)
© 2010 Nature America, Inc. All rights reserved.