Skewed and multimodal lead-time demand (LTD) distributions can be present in supply chains. Under these conditions, logistics managers find that conventional approaches, which assume standard LTD distribution shapes, yield unreliable estimates of a single item’s reorder point (ROP) in relation to meeting fill-rate targets under a continuous review inventory policy. Furthermore, logistics managers have limited data available from which to determine the true shape of the underlying LTD distribution. In this study, I present a non-parametric bootstrap approach to set a least-biased estimate of the ROP. In addition, to deriving bootstrap expressions for non-standard LTD shapes, I derive expressions for standard distribution shapes to estimate the ROP across a range of fill rates. Thus, eliminating the double counting of stockouts with conventional fill-rate measures. Using Monte Carlo simulation experiments, I show that in comparison to the conventional state-of-the-art approaches the non-parametric bootstrap approach yields the least-biased ROP estimates robust to both the shape of the LTD distribution and the sample size. This research offers more accurate ROP estimators, which can serve as a basis for expanding the scope of future inventory management research and enabling logistics managers to reduce inventory while maintaining and even improving fill rates.