Anthropogenic climate change is predicted to cause spatial and temporal shifts in precipitation patterns. These may be apparent in changes to the annual cycle of zonal mean precipitation (P). We show that trends in the amplitude and phase of the P annual cycle in two long-term, global satellite datasets are broadly similar. Model-derived “fingerprints” of externally forced changes to the amplitude and phase of the P seasonal cycle, combined with these observations, enable a formal detection and attribution analysis. Contrary to previous literature, we find that observed amplitude changes are incompatible with both model estimates of internal variability and the response to external forcing. This mismatch between observed and predicted amplitude changes is largely attributable to the sustained La Niña-like conditions that characterize the recent slowdown in the rise of the global mean temperature. However, observed changes to the timing of the wet season onset do not seem to be driven by the recent hiatus. These changes are compatible with model estimates of forced changes, incompatible with estimates of internal variability, and may show the emergence of an externally forced signal.