This paper addresses the worst-case behavior of the LPT (Longest Processing Time first) scheduling heuristic when applied to processors of different speeds. It is shown that, for this model, LPT’s worst-case performance bound lies in the range [1.512, 1.583]. From this we can conclude that LPT is inferior to the MULTIFIT algorithm, which is known to possess a comparable time complexity and has been shown [1] to yield a worst-case bound of less than 1.4. Additionally, for processors of identical speeds, “parameterized” results are obtained which depend on the ratio of the longest task to the overall finish time, in a fashion which is similar in spirit to parameterized results dependent on the number of tasks assigned to a latest-finishing processor [2].