The practice of applying large amounts of P to agricultural land over several previous decades, particularly in regions with intensive livestock production, has resulted in P accumulation in soils with an increased risk of P losses into water bodies and thus of eutrophication. This work is aimed at analyzing the spatiotemporal changes in extractable soil P content at the regional level by means of a soil test database and then comparing results with independent data from agricultural censuses. A total of 228 079 soil P-test values [Dyer method: 1:5 soil-citric acid solution (20 g/L) ratio] generated between 1980 and 2003 by certified commercial laboratories in Brittany (northwestern France) were integrated into a database. Changes in P were analysed by considering four 4-6 year periods between 1980 and 2003 using cumulative frequencies and summary statistics performed on raw soil test values. Then data were aggregated into discrete entities (canton: administrative entities) and summary statistics, linear regression and spatial distribution were carried out. P balances were measured for the same entities taken from the agricultural censuses of 1979, 1988 and 2000. Over the entire study period, a marked spatial variability was observed with higher P content in the western part of Brittany, as well as a systematic increase in median P content with lower amplitude over the past decade. The mean cantonal soil P surplus accumulated over 24 years amounted to 763 kg P/ha of usable agricultural land. Similar P balances sometimes gave rise to widely different increases in P. Cumulative balances were positively correlated with an increase in soil P (r2 = 0.34).