The purpose of this study was to evaluate the pan evaporation method for scheduling irrigation of a sweet pepper (Capsicum annuum) crop grown on an Oxisol at the University of Puerto Rico Agricultural Experiment Station at Isabela, PR. Irrigation scheduling refers to a procedure in which water is applied to the field according to the water demand by the crop. Ideally, water should be applied at a rate equal to the potential evapotranspiration rate of the crop minus the effective rainfall. However, when plastic mulch is used to eliminate weed growth, as is common in vegetable production in Puerto Rico, much of the rainfall runs off and is not available to the plants. To minimize the possibility of crop water stress when plastic mulch is used, water should be applied at a rate equal to the crop evapotranspiration rate. If the contribution from rainfall is assumed to be zero, but in fact some rain enters the soil through holes in the plastic, this may produce deep percolation. Deep percolation is undesirable (except as required for salinity control) because it represents a loss of water and fertilizer, and may also contribute to groundwater contamination. The daily pan evaporation-derived evapotranspiration (ETpan) was obtained from the product of pan evaporation (Epan), a pan coefficient (Kp) and the evapotranspiration crop coefficient (Kc). The contribution of water from rainfall was assumed to be zero. However, some rainfall was observed to enter through holes in the plastic mulch where the pepper plants were located. This contribution of water however, was spatially variable and difficult to quantify. Through calibration of a water balance equation, on average twenty five percent of the rainfall was found to enter the soil beds. Evaluation of the pan method for scheduling irrigation was based on comparison of ETpan with the Penman-Monteith-based evapotranspiration (ETC), estimates of deep percolation, measured vertical hydraulic gradients, and measured soil moisture distribution. Seasonal estimates of ETpan and ETC were 350 mm and 401 mm, respectively. The estimated seasonal deep percolation was 60.9 mm. A simulated irrigation schedule using the Penman-Monteith method resulted in even greater seasonal deep percolation (127.7 mm). Vertical hydraulic gradients were observed to be downward throughout a significant portion of the season, and observed moisture content distributions below the root zone clearly indicated that deep percolation was occurring.