A method is proposed for modeling development of an insect population that passes through a series of discrete developmental stages. This model, including a temperature-dependent development rate, can be estimated directly from field data or can be estimated from laboratory data taken at constant or fluctuating temperatures. The temperature-dependent development rate can have a nonlinear form, and effects on development by environmental factors other than temperature can also be estimated and included. In addition, the method allows for variability among individuals of the population in terms of life-stage duration and allows for correlations among various stage durations. Survival is assumed to be a function of time, but need not be estimated. The proposed method for estimating development of an insect population was applied to spruce budworm,Choristoneura fumiferana(Clemens), field data collected from 23 various location/year combinations in New Brunswick and Maine. The estimated phenology model that best fit the field data included effects of region, host species, calendar date, a linear effect of temperature in New Brunswick, and a nonlinear temperature effect in Maine. The nonlinear development rate estimated for the Maine data set was found to be comparable with development rates in the laboratory. On both the regression data set and an independent data set, the selected model performed better than either a model in which time was measured in calendar-days and better than a model in which time was measured in degree-days.