Abstract Experiments were conducted to clarify the optimal dose and timing of single applications of Ulocladium atrum for control of gray mold on pot rose and pelargonium under conditions similar to those found in commercial greenhouses. In general, application of the antagonist 1-d prior to infestation with conidia of Botrytis cinerea was more effective than application after the infection period. For pot rose, the minimum effective dose when compared to the fungicide fenhexamid was 10 6 conidia ml −1 and persisted for up to 21-d in direct comparison to the fungicide. The effect of pre-emptive application of the antagonist, while significant, was smaller and persisted for a shorter time on pelargonium; when compared to the fungicide, U. atrum was ineffective; however, high application rates and pre-emptive application were more effective than lower application rates or application subsequent to the infection.