The timing of annual cycle events (i.e. phenology) shows evidence for strong responses to climate change in many organisms. However, measuring phenology is associated with several challenges and biases. Dates of first emergence or arrival are strongly affected by population size and observation effort. With densely sampled data within the season, empirical quantiles are a much better option, but may still be noisy or biased if data is truncated, irregularly sampled, or detection depends strongly on weather. A way to deal with these problems is statistical fitting of phenological functions, possibly with weather covariates and appropriate restrictions on the parameters (e.g. Bayesian priors). I claim that using full phenological functions provides better answers to many biological questions relating to phenological change, such as the match–mismatch between a prey and a predatory species.