A new method developed for estimating time-varying survival rates was illustrated by examples with numerical data from a population study onListronotus oregonensis(Le Conte). The conformity between field population and simulated larval population ofL. oregonensissuggested that the simulated results represented the true egg survival rate dynamics in the field. Sensitivity analysis demonstrated that this method was not strongly affected by sampling errors or errors from the model parameters. It was adequate for simulating time-varying survival rates. In field experiments,L. oregonensispopulation density was lower and its peaks were later with later sowing of carrots. The mean survival rates calculated by the new method and by Southwood's area under the curve method were similar. The mean egg survival rates were 0.258 in 1987 and 0.382 in 1988. The egg survival rates in 1987 and 1988 followed a similar pattern in plots sowed at different dates, and there was no obvious relation between the egg survival rates and different sowing dates, although the initial egg population density decreased greatly with the later sowing dates. Egg mortality was density-dependent with no obvious time delay. Average larval survival rates for several sowing dates were 0.295 in 1987 and 0.569 in 1988. Larval mortality was generally density independent, whereas pupal mortality was density-dependent.