Earlier research has extensively examined freight mode and shipment weight dimensions. However, freight destination behavior at a high resolution has received scant attention. In our study, we attempt to address the limited research on destination decision processes and develop a latent segmentation-based approach that accommodates mode and destination choices in a unified framework. The proposed approach postulates that these two choices are actually sequential in nature with an infinitesimally small time gap between them. However, the actual sequence (mode-destination or destination-mode) is unknown to us. Thus, a probabilistic model that can accommodate for the two choice sequences within a single framework is proposed. The latent segmentation framework probabilistically assigns the decision maker to the two sequences. In the Mode first – Destination second (MD) sequence, the destination choice model is calibrated with choice alternatives customized to the chosen mode. In the Destination first – Mode second (DM) sequence, the destination model is calibrated without any mode information as mode is unknown to the decision maker. In the study, we used 2012 US Commodity Flow Survey (CFS) data. We found that the latent segmentation-based sequence model outperformed the independent sequence models (MD and DM). The validation exercise also confirmed the superiority of the proposed framework. Finally, an elasticity analysis is conducted to demonstrate the applicability of the proposed model system.