The role of cowpea (Vigna unguiculata) debris as a primary source of inoculum for Sphaceloma sp., the pathogen of cowpea scab, was studied in field experiments. Three fields were selected in 1993 and three in 1994, in which cowpea had been grown 1, 2 or 3 years previously as part of a crop rotation. Polyethylene mulch was spread over the soil to prevent soil/debris splash in half of the plots. No scab symptoms were observed on the primary leaves. It took an average of 25 days for primary symptoms to be observed in each field, irrespective of mulching. Mulching had a significant effect on disease severity only in the two sites where cowpeas were last grown a year before the trial (sites 1a and 1b). In both years, interaction between time (days after sowing) and site (fields in which cowpea had been grown 1-3 years earlier) had a significant effect on disease incidence while in 1994, interaction between time, site and mulching was also significant (P < 0·05). Higher disease incidence was observed in 1994 than in 1993. In all fields, there were increases in disease incidence over time. Rain splashing may have contributed to higher disease incidence in plots without mulch. The presence of scab in the mulched plots in fields last sown to cowpea 3 years before the trial suggests that the pathogen may survive on infected cowpea debris, which acts as one of the sources of primary inoculum. Hence longer periods of crop rotation with nonhosts may be required to control the disease.