From 2001 to 2004, 252 fifty-plant samples were collected from commercial soybean, Glycine max L., fields in three townships (93-km² area) in Illinois. Townships were sampled every 3 wk from late June or early July when aphids (Aphis glycines Matsumura) first invaded the townships to early August. We used linear regression of 18 mean township field densities to calibrate several simple models to predict the change in aphid population density in a township from one sampling date to the next. The best exponential model for the complete data set has an r² = 0.54, Y2 = Y1exp(0.09659 × DAY), where Y1 and Y2 are the first and second samples of aphids separated by a 3-wk period (the number of days, DAY). Our intrinsic rate of increase for the population is much lower than rates calculated in other studies. The best single-variable linear model has an r² = 0.88, Y2 = Y1 0.1084 × Y1xDAY. The latter model indicates the value of including monitoring data in the prediction. The best two-variable model has an R² = 0.98, Y2 = Y1 0.08136 × Y1xDAY 0.000080 × N1²xDAY, where N1²xDAY is the interaction term for initial, squared, sample density of the season multiplied by the number of days between samples. The latter two models indicate that the change in the population density is greater for more dense populations. Degree-days were generally inferior to days as the time component in the simple models.