This paper examines the spillover effects of television brand advertising on consumer demand for carbonated soft drinks using a nested logit model and household purchase and advertising data from nine U.S. cities over a three year period. Spillover effects are modeled using the conventional linear advertising production function with brand and company advertising, which assumes perfect substitution and constant returns to scale, and the results are compared to those attained via a constant elasticity of substitution (CES) model. Empirical results indicate that brand advertising spillover effects have a positive and significant impact on the demand for all brands belonging to the same company, highlighting the importance of accounting for spillover effects in demand models. The CES advertising production function outperforms the linear form, indicating decreasing returns to advertising goodwill and imperfect substitutes between brand advertising and company advertising.