This study explores the determinants of food expenditures in northern Ghana’s rural households, using a survey data collected in 2010 in the vicinity of Tamale, the capital of the Northern Region. Three estimation methods (OLS, OLS with robust error, and WLS) are used in empirical models to address the possible heteroscedasticity. Models indicate that socio-demographic factors such as income, owning a tractor, age, and household composition are important factors in determining food expenditure. Similarly, farm features such as cultivation of staple or cash crops, the field size of groundnuts, as well as buying dry goods in bulk are also found to be major determinants. Results provide useful information for both private and public sector decision makers, while supplying ample evidence of the importance of estimation method selection to generate most accurate quantified effects of individual explanatory variables on food expenditure.