This study estimated Indonesian households' Multidimensional Energy Poverty Index (MEPI) using Alkire-Foster's multidimensional poverty concept to capture the incidence, intensity, decomposition, and changes over time. The study used monetary and non-monetary variables to identify the availability, accessibility, affordability, consumption, and deprivation of modern energy services; and compared existing affordability indicators. Redundancy, robustness, and sensitivity tests were conducted with three weighting schemes and deprivation cut-offs. The study decomposed and determined household-head (HH) socio-economic, demographic, and geographic factors for MEP using the Logit, Probit, Tobit, and Heckman Selection models. The results show that the low-income and high cost (LIHC) was the most robust affordability indicator, followed by the ten percent rule (TPR). The complement-frequency weighting scheme gave the smallest and most robust MEPI compared to equal and principal component analysis (PCA) weighting. Three alternative deprivation cut-offs can show households as “vulnerable,” “moderately,” or “severely” energy poor. The MEP incidence decreased, but its intensity remained high and increased. Energy-poor households were averagely deprived of 55–60 % of all weighted indicators. The lack of modern cooking services was the primary cause. MEPI differed by geographical location and HH gender, education, business field, and employment status. Policies that boost education levels, raise household income, and increase the availability, accessibility, and affordability of modern cooking technology in the rural, hinterland, or non-coastal forested locations in the eastern islands of Indonesia may minimize the number of households experiencing MEP.