Autocorrelation is a concern when looking for patterns in time-ordered data, such as residuals, X and MR charts and EWMA charts. If autocorrelation exists, it implies that variations in values over time are not random, that is, if the nth value in the sample is high, the (n-1)th value is also likely to be high, etc. In the analysis of a time-ordered chart of Residuals, this implies that the assumption of independence has been violated. Positive autocorrelation exists if the residuals do not change signs as frequently as expected. Negative autocorrelation exists if the Residuals change signs more frequently than expected.
Source: Breyfogle III, Forrest W.(2003). Implementing Six Sigma: Smarter Solutions Using Statistical Methods, 2nd ed., John Wiley & Sons, New York.
NIST Statistics Handbook: Autocorrelation - http://www.itl.nist.gov/div898/handbook/eda/section3/eda35c.htm