The objectives of this article are, first, to provide improved estimates of recent fertility levels and trends in Nepal and, second, to analyse the components of fertility change. The analysis is based on data from Nepal’s 1996 and 2001 Demographic and Health Surveys. Total fertility rates (TFR) are derived by the own-children method. They incorporate additional adjustments to compensate for displacement of births, and they are compared with estimates derived by the birth-history method. Fertility is estimated not only for the whole country but also by urban/rural residence and by woman’s education. The own-children estimates for the whole country indicate that the TFR declined from 4·96 to 4·69 births per woman between the 3-year period preceding the 1996 survey and the 3-year period preceding the 2001 survey. About three-quarters of the decline stems from reductions in age-specific marital fertility rates and about one-quarter from changes in age-specific proportions currently married. Further decomposition of the decline in marital fertility, as measured by births per currently married woman during the 5-year period before each survey, indicates that almost half of the decline in marital fertility is accounted for by changes in population composition by ecological region, development region, urban/rural residence, education, age at first cohabitation with husband, time elapsed since first cohabitation, number of living children at the start of the 5-year period and media exposure. With these variables controlled, another one-third of the decline is accounted for by increase in the proportion sterilized at the start of the 5-year period before each survey.