Selected environmental and life-cycle determinants of change in household composition were examined using longitudinal data. Although cross-sectional analysis would suggest that 12% of a random sample of 276 older community residents had experienced a change in type of household composition over a 30-month period, longitudinal analysis showed that 18% experienced a change in major type of household composition, some types of households being more stable than others. Longlinear analysis showed that these changes were best accounted for by earlier life cycle events that were largely beyond the individual's present control (i.e., by change in marital status, typically death of spouse, and by extent of help available from family and friends). The importance of such family life cycle events in explaining change in household composition was emphasized.