This letter focuses on exploring a new decoding order to resolve the secrecy issue among untrusted nonorthogonal multiple access users. In this context, firstly, we identify the total number of possible decoding orders analytically. Then, we propose a decoding order strategy ensuring positive secrecy rate for all users, and obtain the number of feasible secure decoding orders based on this proposed strategy numerically. Thereafter, we present a suboptimal policy to avoid the computational complexity involved in finding the best secure decoding order. Numerical results illustrate that the suboptimal solution provides a performance gain of about 137% over therelevant benchmark.