Higher nitrogen (N) deposition and altered precipitation patterns are two important global change factors contributing to increasing atmospheric concentrations of greenhouse gas nitrous oxide (N2O). In most experiments manipulating global change factors, wet N deposition is simulated by acute N addition once or a few times per year, rather than chronic low N inputs associated to rainfalls as they occur in the field. Whether N deposition pattern (acute or chronic) matters for soil N2O emissions, and whether altered rainfall frequency and modified N deposition pattern have an interactive effect on soil N2O emissions, are still knowledge gaps. Here we used a full factorial design for a mesocosm experiment in a semi-arid grassland. The experimental factors included rainfall frequency (ambient number of rainfall events; +50% number of rainfall events; and -50% number of rainfall events), and N deposition pattern (no N deposition; acute N deposition with two N inputs; and chronic N deposition co-occurring with rainfall events). Our results showed that altered rainfall frequency had no effect on soil N2O emissions irrespective of N deposition pattern. In contrast, N deposition pattern matters for soil N2O emissions. As compared to no N addition, cumulative N2O emissions during the vegetation growing season were significantly increased by 36% for chronic N deposition, but not significantly for acute N deposition. The temporal dynamics of N2O fluxes also varied between acute and chronic N deposition. These findings suggest that acute N addition may underestimate the actual N deposition effect on N2O emissions in grasslands, and that N deposition pattern should be considered when building models of ecosystem-climate feedback.