In this paper, we consider three blind channel estimation methods. Cross-relation (CR), subspace (SS) and least squares smoothing (J-LSS) methods are compared for single-input multi-output (SIMO) systems. In contrast to the previous works, we evaluate the practical MSE performances of these methods for short data lengths and random channels when the number of channels is greater than two. Some previously unknown characteristics of these methods are presented. A novel method for blind channel order estimation is presented and it is integrated to the SS and CR methods. The performance of this approach is compared with different methods. It is shown that the new approach has significantly better performance and it is robust to a variety of factors in practical applications.