Purpose - This paper empirically assesses the determinants of conditional stock index autocorrelation with particular emphasis on the impact of return volatility that are theoretically linked through the behaviour of feedback traders. Design/methodology/approach - The S&P 100, 500 and the NASDAQ 100 index are considered and volatility in each series is captured using option-implied estimates taken from the Chicago Board Options Exchange. A seemingly unrelated regression approach is used in which trading volume and volatility are simultaneously modelled. Findings - The results of this study suggest that low or even negative return autocorrelations are more likely in situations where: return volatility is high; price falls by a large amount; traded stock volumes are high; and the economy is in a recessionary phase. Research limitations/implications - The results confirm that previous related work showing a link between autocorrelation and volatility is not induced by a mechanical relation. Practical implications - Usage of endogenously determined volatility measures in this area of the literature is justified. Originality/value - This study provides a robustness test of the autocorrelation/volatility relation, as well as a further exploration of the utility inherent in option-implied volatility.