We analyze the performance of the decorrelating multiuser detector for a wireless communication system. Instead of looking at the error probability for each user, we focus on the probability distribution of the number of correctly decoded users at the receiver and the average number of correctly decoded users under the conditions of equicorrelated signatures and perfect power control. The approach is motivated by the fact that the error events for different users are not independent. Numerical results show that when the number of users is large, instead of using a set of long signatures with a very low negative cross-correlation, we can use a set of shorter signatures with a reasonably high positive correlation without sacrificing the performance. It is also found that the average number of correctly decoded users increases with increase in signal-to-noise ratio and decreases when the cross-correlation between users becomes more negative or more positive.