A crop growth model of the Food and Agricultural Organization of the United Nations (FAO) adapted for estimating yield potential of annual crops in Canada was modified to provide climate-based estimates of perennial forage yields (FORYLD). Subroutines were developed to estimate average cutting schedules and growth periods for legumes (alfalfa) and grasses (timothy, bromegrass, wheatgrass) from climatic normals data. Model estimates of cutting schedules and potential (constraint-free) yields without moisture stress (B(y)) were in good agreement with observed values. A moisture stress factor (MSF) was used to estimate dry matter yields (B(ya)) under rainfed conditions (i.e. B(ya) = MSF x B(y)). The MSF was determined from the soil moisture deficit-surplus (D), which was based on available water content in the soil at the start of the growth period, rainfall and potential evapotranspiration over the growth period. In a test of individual cuts, using dependent data, a high percentage of cases (82%) had estimated values for B(ya) within 1 t ha-1 of observed values for alfalfa and timothy. In independent data tests of wheatgrasses, 95% and 58% of estimated cases were within 1.0 t ha-1 and 0.5 t ha-1 of observed values, respectively. Average constraint-free yield (B(y)) and dry matter yields under rainfed conditions (B(ya)) were estimated for all soil map units on the Soils of Canada map, using FORYLD with 30-year averages of monthly climatic data. Estimates of B(y) for both alfalfa and grasses ranged from zero potential in northern areas to more than 18 t ha- 1 in the most climatically favoured areas (southern Quebec, lower coastal areas in British Columbia, southern Alberta), based on a maximum of three cuts of alfalfa and two cuts of grasses. Estimated values of B(ya) for alfalfa in areas where By was greater than zero ranged from less than 2 t ha-1 in the semi-arid regions of the prairies and parts of the interior of British Columbia, to more than 14 t ha-1 in the most climatically favourable areas of southern Ontario and Quebec and the lower coastal region of British Columbia. The results provide a reasonable basis for making comparisons of forage production capability between regions. Additional consideration of stand persistence and winter injury is needed if yield potential data are to be used for management decisions on forages.