Control charts have played a key role in industrial process control since the 1930's. When dealing with a quality characteristic of numerical measurements, the X-bar and R charts are generally used to control the process mean and variability, respectively. Traditionally, 3-sigma control limits are constructed in both charts. The numerical measurements are usually assumed to be normally distributed. However, this assumption may not be true. Although the sampling distribution of X-bar approaches normality as the sample size is sufficiently large, the sampling distribution of R is always asymmetrical, no matter what the underlying distribution is. Therefore, if the 3-sigma limits are still used in the R chart, the control ability may be reduced. This paper employs the Burr distribution to construct the control limits for an R chart under non-normality and then, compares the non-normal control limits with the traditional 3-sigma limits based on the average run length of an R chart.