We propose a framework for modelling plant reproductive data. Several statistical issues can arise in the analysis of reproductive data and this paper develops a framework for dealing with them. The relationship between reproductive output and plant biomass regularly follows a log–log allometric regression. Frequently, a number of plants do not reproduce and the corresponding zero reproductive values do not fall easily within this standard regression framework. Truncated regression allows zero values to be incorporated appropriately in the allometric relationship. We also propose a mixture-model method to deal with outlier values that do not follow the allometric relationship, for example large plants that have zero reproductive output values. Reproductive data from plants grown together in pots or in field plots may not be independent and this dependence should be dealt with in each of the above analyses. We illustrate our method using either generated data or data from an experiment examining reproductive output in Sinapis arvensis and provide the programming tools used. We test our methods and compare them with widely used alternatives using simulation studies. These studies validate the use of our proposed approach and show that some of the alternatives produce seriously biased estimates of model parameters. We also present a general graphical aid to assist the selection of the appropriate method to analyse plant reproductive data.