Clean, diversified and sustainable household energy sources for cooking is essential in order to maintain worthy health for women and children and also improving the energy security of people in the developing countries. Yet, the understanding of household energy dynamics and information remains unclear. This necessitates investigation of transition pathways towards diversification, sustainable and modern household energies. The main objective of this research was to model household energy utilization, changing behaviours and diversification using Structural Equation Modelling (SEM). The specific objectives included: determinants of household energy utilization and changing behaviours; the effects of renewable energy and accessibility on energy utilization, changing behaviour and household diversification of energy sources and finally modeling of the effects of moderators and mediators on the household energy sources diversification. The research was carried out in the counties of Bungoma and Uasin Gishu. Random sampling technique was used to select 640 households from a target household of 663,739 and data was collected using a structured questionnaire. The data was analyzed using AMOS version 23 to achieve the first three objectives. Bootstrapping method was utilized to validate mediation and moderation models. The results showed that firewood is still the most common energy resource used for cooking in both rural and peri urban areas as evidenced by responses of 87.5% and 72.4%, respectively. The use of LPG (26 to 42%), charcoal (39.4% to 53.8%) and kerosene (14.3% to 17.3%) for cooking was found to increase as one moves from rural to peri-urban and vice versa for agricultural residues (12.3% to 5.3%). Biogas uptake still represents a small fraction (11.4 to 14.6%) of the energy mix at local level. The use of solar for lighting showed reduction as one move from rural to peri urban (44.8% to 39.6%) and vice versa for kerosene and electricity. SEM analysis found that factors such as education level, income, residential status, peri urbanization, house size, house composition, age and gender of the household head influence the changing behaviours and diversification among households both for cooking and lighting. Biogas users realized time saving of 1hour 36 minutes on average per household daily with financial saving of KES 2,557 per month as compared to firewood users. In addition, biogas indicated negative association with the use of conventional household energy sources for cooking fuels. Consequently, accessibility increased household fuel utilization and diversification. Interestingly, LPG (Path coefficient () = 0.461, critical ratio (C.R) = 15.204) followed by biogas ( = 0.333, C.R = 11.738) revealed to be the most important contributor to household diversification. The mediating effects of peri urbanization improved the household utilization of charcoal ( = 0.01, C.R = 6.72) kerosene ( = 0.04), LPG ( = 0.01), and conversely for firewood ( = - 0.013, C.R = 8.72) and agricultural residues ( = - 0.01). With income as an independent variable and education as a moderator; number of cars ( = 0.21), peri urbanization ( = 0.01), household size ( = 0.0397), residential status ( = - 0.0396), and gender ( = - 0.104) revealed mediating effects on the household energy diversification. According to bootstrapping reliability test, the limit for Bollen-Stine bootstrap is < 0. 12. In conclusion, household attributes have direct, moderating and mediating effects on the household energy utilization, changing behaviour and diversification. This study showed that household energy changing behaviour and diversification in Kenya are affected by moderating and mediating factors such as peri urbanization, cars among others. This study puts forward the need for policymakers and energy planners in Kenya and other developing countries to improve accessibility (supply and distance) of sustainable fuels and create awareness about the harmful effect of using dirty fuel at early stage through education curriculum, seminars and workshops.