Databases amassed from the literature were used to predict feed intake by lactating, Angora, growing and mature goats, using 221, 54, 282 and 99 treatment means, respectively. One prediction approach was based on a calculated constant overall efficiency of ME utilization (k) considering biotype (meat, ≥50% Boer; dairy; indigenous; Angora), BW (kg; all goats), 4% fat-corrected milk (FCM, kg; lactating), BW change or ADG (kg; lactating, growing and mature), dietary ME concentration (MEC, MJ/kg DM; all goats), tissue gain (TG, kg; Angora) and clean mohair fiber gain (FG, kg; Angora). For lactating goats, assumptions included efficiency of ME utilization for maintenance and activity: 0.503+(0.019×MEC); efficiency of ME use for gain (kg): 0.75; efficiency of use of mobilized ME for lactation: 0.84; efficiency of use of dietary ME for lactation: 0.589; tissue energy concentration (TEC): 23.9MJ/kg; ME requirement for maintenance and stall or pen activity (MEmREQ): 0.5013 and 0.4227MJ/kg BW0.75 for dairy and other goats, respectively; and all mobilized tissue energy used for lactation. After removing observations with residuals greater than 1.5×root mean square error (RMSE), k was 0.653 (S.E.=0.0014). Predicted DM intake (DMIP) including an adjustment (DMIAP) for the ratio of ADG:FCM (ADGFCM) was: DMI=0.0964(S.E.=0.0704)+(0.9334(S.E.=0.9314)×DMIP)−(0.1237(S.E.=0.05923)×ADGFCM) (R2=0.84;RMSE=0.2187;n=191). Mean k, estimated from a random development data set, resulted in unbiased prediction of intake for an evaluation data set without observations removed. Assumptions for Angora goats that differed from lactating goats were efficiency of ME use for tissue gain (TG; kg/day): 0.006+(0.0423×MEC); efficiency of use of ME (dietary and mobilized tissue) for clean fiber gain (FG): 0.151; TEC=4.972+(0.3274×kgBW); MEm: 0.473MJ/kgBW0.75; ME used for FG: FG×157MJ/kg; and all mobilized tissue energy used for FG. Mean k for Angora goats was 0.525 (S.E.=0.0112), and prediction accuracy was improved by adjusting for dietary CP concentration (PTCP, % DM): DMI=−0.1607(S.E.=0.11430)+(0.8227(S.E.=0.10851)×DMIP)+(0.0199(S.E.=0.00697)×PTCP)(R2=0.65;RMSE=0.1239;n=54). Assumptions for growing goats included: kg: 0.006+(0.0423×MEC); efficiency of use of mobilized tissue energy for maintenance: km; and MEmREQ: 0.489, 0.580 and 0.489MJ/kgBW0.75 for meat, dairy and indigenous goats, respectively. After removing observations with residuals greater than 2×RMSE, k was 0.634(S.E.=0.0020). Prediction accuracy was improved by adjusting for ratios of ADG to BW (ADGBW), BW0.75 (ADGMBW) and ADGMBW2: DMI=−0.0047(S.E.=0.1854)+(0.9637(S.E.=0.04928)×DMIP)−(70.27(S.E.=23.534)×ADGBW)+(38.71(S.E.=12.224)×ADGMBW)−(243.4(S.E.=121.73)×ADGMBW2)(R2=0.88;RMSE=0.1030;n=266). Mean k estimated from a random development data set resulted in unbiased prediction of intake for an evaluation data set without observations removed. Assumptions for mature goats were the same as those for growing goats except for a MEmREQ of 0.462MJ/kgBW0.75. k was 0.632(S.E.=0.00448), and prediction accuracy was improved by adjusting for PTCP, ADGBW and ADGMBW: DMI=−0.1241(S.E.=0.07374)+(0.7915(S.E.=0.06911)×DMIP)+(0.0214(S.E.=0.00381)×PTCP)−(535.2(S.E.=66.35)×ADGBW)+(247.3(S.E.=29.53)×ADGMBW)(R2=0.85;RMSE=0.1537;n=99). Because of the relatively large number of observations in this study, these methods should be useful for predicting voluntary intake of different diets by a variety of goats in or near thermoneutral conditions fed in pens or stalls.