Groups of 25 healthy dogwood seedlings were exposed for 2 wk to naturally occurring inoculum under mature, diseased trees at 2-wk intervals for three growing seasons, After exposure, seedlings were placed in an incubation room and supplied with trickle irrigation and fluorescent lighting for 2 wk. Following incubation, percent leaf area infected (LAI) was estimated visually. In 1989, LAI remained less than 5% until June. However, consistent rainfall throughout the summer created conditions conducive to infection of seedling groups exposed from 6 June through September, and LAI ranged from 11 to 47% during that period. In 1990 and 1991, LAI was less than 5% until early May, and heavy infection began in mid-May. Midsummer droughts reduced LAI to less than 5%. LAI increased with renewed rainfall but dropped below 10% for the remainder of the season beginning in mid-July 1990 and in September 1991. Numerous secondary infection cycles occurred in each of the 3 yr. Stepwise regression analysis showed that 34% of the variance in LAI was explained by 2-wk rainfall total and 17% was explained by the LAI of the previous seedling group. This supports the hypothesis that secondary infection cycles in southwestern North Carolina depend on consistently recurring rainfall and inoculum buildup.