Abstract: The performance of direct sequence spread spectrum(DSSS) communication systems in the presence of narrowband interference is studied when decision feedback filters are employed for estimating and suppressing the interference. Simple upper bounds of signal to noise ratio(SNR) improvement factor are derived for the suppression of a narrowband first-order autoregressive process and analytical conclusions corroborated by numerical examples. To obtain the same performance, the orders needed from decision feedback filters proposed are much lower than linear prediction filters, making hardware implementation more tractable. Both analysis results and numerical results by Monte Carlo simulation showed that decision feedback filters(DFF) outperform linear prediction filters by comparison.