Abstract. Much of population biology is still based on deterministic approaches, and even many stochastic approaches make a number of simplifying assumptions. In this talk, I will cover two areas: (1) highly replicated experiments in the laboratory using flour beetles and accompanying models and what they tell us about stochasticity and predictability in population biology, and (2) some surprising aspects arising from simple models that take into account autocorrelation, demography and other specific problems in population biology. The first part of the talk on flour beetle experiments will highlight the importance of understanding sources of stochasticity and the role played by careful models. The second part of the talk will emphasize the consequences of simplifying assumptions and the importance of time scales in the analysis of ecological models.