Conservation action is usually triggered by detecting trends in species’ population size, geographical range, or occupancy (proportion of sites occupied). Robust estimates of these metrics are often required by policy makers and practitioners, yet many species lack dedicated monitoring schemes. An alternative source of data for trend estimation is provided by biological records, i.e., species presence information. In the UK, there are millions of such records, but biological trend assessments are often hindered by biases caused by the unstructured way in which they are collected. Recent advances in occupancy modelling that account for changes in survey effort and detectability over time mean that robust occupancy trends can now be estimated from these records. By grouping mammal species into survey assemblages — species likely to be recorded at the same time — and applying occupancy models, this study provides estimates of long-term (1970 to 2016) occupancy trends for 37 terrestrial mammal species from the UK. The inter-annual occupancy growth rates for these species ranged from -4.26% to 11.25%. This information was used to classify two species as strongly decreasing, five as decreasing, 12 as no change, 11 as increasing and seven as strongly increasing. Viewing the survey assemblages as a whole, the occupancy growth rates for small mammals were, on average, decreasing (-0.8% SD 1.57), whereas bats and deer (0.9% SD 1.30) were increasing (3.8% SD 3.25; 0.9% SD 1.30 respectively), and mid-sized mammals were stable (-0.3 SD 1.72). These results contribute much-needed information on a number of data deficient species, and provide evidence for prioritising conservation action.