In New Zealand, phosphate (P) fertilisers used in agriculture are the main sources of the potentially toxic elements cadmium (Cd) and uranium (U), which occur as unwanted contaminants. New Zealand is developing draft soil guideline values (SGV) for maximum concentrations of Cd. To assess when soils under pasture for sheep production might reach a particular SGV, we analysed archived soil samples from a 23 yr P fertiliser trial. The pasture sites were at Whatawhata, North Island, New Zealand, and had received P fertiliser at the rates of 0, 30, 50 and 100 kg P ha⁻¹ yr⁻¹. From 1983 to 1989, P was applied as single superphosphate, from 1989 to 2006, P was applied as triple superphosphate. Soils from replicate paddocks were sampled annually to a depth of 75 mm on easy (10–20°) and steep (30–40°) slope classes. Total P, Cd and U were analysed by ICP-MS after acid digestion. Data were analysed by fitting trend lines using linear mixed models for two slope classes and for two sampling periods 1983–1989 and 1989–2006 when the soil sampling method and fertiliser type had been changed. The changes in total P, Cd and U were directly related to the type and amount of P fertiliser applied, the control treatment showed no significant change in P, Cd or U. At 50 and 100 kg P ha⁻¹ yr⁻¹ there were generally linear increases in total P and total U, and the same trend line applied to both time periods, but the rate of increase in P was greater on the easy slope class. For Cd, a “broken stick” model was needed to explain the data. Pre-1989, Cd increased in the 50 and 100 kg P ha⁻¹ yr⁻¹ treatment (0.036–0.045 mg kg⁻¹ yr⁻¹, respectively): post 1988 the rate of increase declined markedly on those two treatments (0.005–0.015 mg kg⁻¹ yr⁻¹, respectively), and declined absolutely in the 30 kg P ha⁻¹ yr⁻¹ treatments. The maximum content of Cd was in the 100 kg P ha⁻¹ yr⁻¹ treatment which reached 0.931 mg Cd kg⁻¹ on the easy slope. For U there were steady linear increases for the 30, 50 and 100 kg P ha⁻¹ treatments, and no significant difference between the steep and easy slopes, nor the two sampling periods, the maximum concentration obtained was 2.80 mg U kg⁻¹ on the 100 kg P ha⁻¹ treatment. The results suggest that at rates of P fertiliser likely to be applied to hill farms (<50 kg P ha⁻¹ yr⁻¹ ), and using P fertiliser with low Cd content, then the Cd concentration in this soil will never reach a SGV of 1 mg kg⁻¹.