Abstract A trend of increasing rainfall over much of north and northwest Australia over recent decades has contrasted with decreases over much of the rest of the continent. The increases have occurred during the summer months when the rainy season is dominated by the Australian monsoon but is also affected by other events such as tropical cyclones, Madden–Julian oscillations, and sporadic thunderstorms. The problem of diagnosing these trends is considered in terms of changes in the timing of the rainy season. While numerous definitions for rainy/monsoon season onset exist, most are designed to be useful in a predictive sense and can be limited in their application to diagnostic studies, particularly when they involve predetermined threshold amounts. Here the authors define indices, based on daily rainfall observations, that provide relatively simple, robust descriptions of each rainy season at any location. These are calculated using gridded daily rainfall data throughout the northern Australian tropics and also for selected stations. The results indicate that the trends in summer rainfall totals over the period from 1950 to 2005 appear to be mainly the result of similar trends in average intensity. Furthermore, the links between the September–October average Southern Oscillation index indicate that ENSO events affect season duration rather than average intensity. Because duration and average intensity are derived as independent features of each season, it is argued that the trends in rainfall totals are largely unrelated to trends in ENSO and most likely reflect the influence of other factors. Finally, diagnosing these features of the rainy season provides a basis for assessing the confidence one can attach to different climate model projections of changes to rainfall.