We determined decadal changes in soil carbon (C) and nitrogen (N) due to different irrigation regimes and phosphorus fertilization of pastures. Archived soil samples (0–75 mm) collected annually from two long-term trials in New Zealand were analyzed for %C and %N from three P input treatments (ranging from 0 to 376 kg superphosphate ha−1 yr−1, 1952–2009) and three irrigation treatments (unirrigated and irrigated when soil moisture content fell below either 10 or 20%, 1959–2002). In the fertilizer trial, soil C increased linearly from 2.7 to 4.2%, and there was no difference in rates of increase in C between treatments, despite much greater aboveground production when P was added. This lack of difference was attributed to higher stocking rates on treatments with higher production, and to the possibility that root inputs (which differed less between treatments) was a more important control of soil C accumulation. Nitrogen (%) was lower in the unfertilized than fertilized treatments due to lower clover N fixation, which was constrained by P availability. Soil C (%) was significantly greater in the unirrigated treatment than the irrigated treatments throughout the trial. Aboveground production was much greater in the irrigated than dryland treatment but root biomass was lower. Irrigation must have increased C and N losses, possibly via increased respiration rates during seasonally dry periods. Our study showed that P fertilizer application did not result in an increase in surface soil C and that flood irrigation resulted in a constrained increase in surface soil C content.