In this paper, we study an enhanced subspace based approach for the mitigation of multiple access interference (MAI) in direct-sequence code-division multiple-access (DS-CDMA) systems over frequency-selective channels. Blind multiuser detection based on signal subspace estimation is of special interest in mitigating MAI in CDMA systems since it is impractical to assume perfect knowledge of parameters such as spreading codes, time delays and amplitudes of all the users in a rapidly changing mobile environment. We develop a new blind multiuser detection scheme which only needs the priori knowledge of the signature waveform and timing of the user of interest. By exploiting the improper nature of multiple access interference (MAI) and intersymbol interference (ISI), the enhanced detector shows clear superiority to the conventional subspace-based blind multiuser detector. The performance advantages are shown to be more obvious in heavily loaded systems when the number of active users is large.