Multinomial logistic regression models are popular in multicategory classification analysis, but existing models suffer several intrinsic drawbacks. In particular, the parameters could not be uniquely determined because of the over-specification. Some additional constraints have been imposed to refine the model but such modifications can be inefficient and complicated. In this paper, we propose a novel and efficient simplex-based multinomial logistic regression technique, seamlessly connecting binomial and multinomial cases under a unified framework. Compared with the existing models, our model has less parameters and is free of any constraints, which can be efficiently solved via the Fisher scoring algorithm. In addition, the proposed model enjoys some theoretical advantages including Fisher consistency and sharp comparison inequality. Under mild conditions, we establish the asymptotical normality and convergence for the new model even when the numbers of categories and covariates increase with the sample size. The proposed framework is illustrated by extensive simulations and real applications.