An enrichment method was used to monitor Erwinia carotovora in soil or the rhizosphere of different crops and weeds in 17 fields with different cropping histories on three farms. The bacteria were detected in all fields not cropped with potatoes, although not consistently, and the mean annual frequency of detection was generally low (< 10%). Fields in which potatoes were grown were extensively contaminated after harvest in September but contamination declined over the winter to very low levels by early summer in the following year. Contamination level tended to rise in some fields without potatoes regardless of their cropping history but for only a short time during autumn and winter. The bacteria were no more frequent in rhizosphere soil of any of the weeds or crops examined, with the exception of brassicas, than in bare soil. In fields where more than 16 months had elapsed since cropping with potatoes, 91% of erwinia isolates obtained were E. carotovora subsp. carotovora, the remainder being E. carotovora subsp. atroseptica. The bacteria were shortlived in soil and in the rhizospheres of inoculated field and pot grown crop and weed plants. Longevity was greater in dry (10% moisture) than in wet (21% moisture) soil and decreased as temperatures rose, particularly above 25°C. Survival was best in association with brassica plants, moderate on grasses and cereals, and least on potatoes and weeds. E.c. carotovora survived better than E.c. atroseptica. Because survival of the bacteria in soil is apparently restricted, their presence in fields could be attributed to recurrent introductions from different sources.