Multiplicative and mediating variables are combined with a demographic variable, in non-linear multiple regression models to assess the effect of population growth on agricultural land use in two agro-ecological zones of Ghana. The paper uses data from a retrospective household survey (conducted among 1568 farmers in 504 households in 24 communities), population census reports of Ghana, for 1960, 1970, 1984 and 2000, agricultural census data for 1992 and 2000 and a land suitability map. Predictions of cropped area required to meet anticipated population growth are computed for 2010, based on multiple regression models and projected populations. Required cropped area is matched with actual arable area that would be available, based on a land suitability analysis. Predictions show that three and two districts in the dry and derived savannah zones, would, respectively, experience agricultural land shortfall. The rest of the districts in the two agro-ecological zones would have agricultural land available to support future population growth.