Density patterns of colony-forming units (CFU) of the biocontrol agent Gliocladium roseum were investigated in raspberry cv. Boyne and cv. Redwing after spore suspensions (10 7 conidia/mL) were applied 2 h before dusk in field plots. Estimated density of G. roseum on leaves of both cultivars was 2 x 10 3 to 5 x 10 3 CFU/cm 2 at 1 h after inoculation. During the next 2 days density of the agent progressively decreased, but at 3-10 days after inoculation it fluctuated around 10-10 2 CFU/cm 2 leaf in 'Boyne' and after or 3-12 days, it fluctuated near 10 3 CFU/cm 2 leaf in 'Redwing'. Density of G. roseum on flowers was 1 x 10 4 to 5 x 10 4 CFU!flower at 1 h after inoculation and declined rapidly in 'Boyne' and slowly in 'Redwing'. At 3-10 and 3-12 days, respectively, the density fluctuated around 70 CFU/flower in 'Boyne' and 800-2300 CFU/flower in 'Redwing'. Germination of G. roseum, observed on stamens of 'Boyne', was sparse (<4-10%) in the field. However, germination potential, estimated in stamens of flowers removed and placed in high humidity, exceeded 80% at 1 h after inoculation, declined slowly at night and rapidly by day at 1-48 h after inoculation, and subsequently ranged from below 6 up to 9%. Loss in germination ability was a key factor in reduced CFU and, from correlative and circumstantial evidence, was favoured by high temperature and irradiance. Death of germinated conidia and endophytic establishment of G. roseum, each predicted on several nights, also contributed presumptively to net reductions in CFU. Sporulation and redistribution of G. roseum potentially contributed to net increases in CFU. Gliocladium roseum completely suppressed sporulation of Botrytis cinerea in leaves and stamens that were challenge-inoculated with a high density of the pathogen at 12-96 h after the antagonist was applied, and by 33-83% in leaves and 23-93% in stamens at subsequent times. Observed density patterns have implications for strategic timing of G. roseum treatments against B. cinerea in raspberry.