The problem of nonlinear signal separation and underdetermined signal separation has received increasing attention in the research of blind signal separation. Few of them can solve the situation where nonlinear and underdetermined characteristics exist simultaneously. In this paper, a new learning algorithm based on Bayesian statistics is proposed to solve the problem of the blind separation of nonlinear and underdetermined mixtures. This paper addresses the Blind Signal Separation (BSS) of post-nonlinearly mixed signals where the number of observations is less than the number of sources. Formal derivation shows that the source signals, mixing matrix and nonlinear functions can be estimated through an iterative technique based on alternate optimization. Simulations have been carried out to demonstrate the effectiveness of the proposed algorithm in separating signals under nonlinear and underdetermined conditions.