We propose a logistic regression model that represents the process of generating subjective probabilities for combined conditionals, including concrete contents. We assume the generation process to consist of two components. The first is the calculation of subjective probabilities for two simple conditionals that have the same consequent (i.e., A → C, B → C), which constitute a combined conditional (i.e., A ∧ B → C). The second component is the semantic interaction between the two antecedents and the consequent. In the proposed model, the subjective probability for a combined conditional is calculated based on the logistic function for the linear combination of the latent strengths for the two simple conditionals involved in the combined conditional. The coefficients of the linear combination represent the semantic interaction between the two antecedents and the consequent. The latent strengths and the coefficients in the model are estimated from subjectively rated probabilities. These estimated parameters are used to classify the combined conditionals according to two types of semantic interaction, which are discussed in terms of the actual semantic content of each combined conditional.