This paper presents a selection methodology and performance analysis of short optimized chaotic spreading sequences for direct sequence code division multiple access (DS-CDMA) communication systems. Binary non-return-to- zero (NRZ) chaotic sequences, generated by Logistic, modified Bernoulli, modified Tent, Gauss, Sine-Circle, Cubic, and Pinchers one-dimensional maps and optimized using high volume compu- tations, are used. Performance of a DS-CDMA system based on selected chaotic spreading sequences is compared to a system without sequence selection under the assumption of perfect chip synchronization. A brief statistical analysis of periodic cross- correlation properties of the selected chaotic sequences, generated by the aforementioned maps, is presented. Moreover, the impact of the chaotic sequence length, the number of users, as well as the impact of the chaotic sequence generation algorithm on the performance of a DS-CDMA system is studied.