Considerable disagreement exists between results reported by various authors for lipid composition and enzyme activity in purified muscle membrane fractions presumed to be sarcolemma, although an explanation for these discrepancies has not been presented. We have prepared muscle light surface membrane fractions of comparable density (1.050--1.120) by a low-salt sucrose method and by an LiBr-KCl extraction procedure and compared them for density profile, total lipid and cholesterol content, protein composition and ATPase activity. In addition, sodium channels characteristic of excitable membranes have been quantitated in each preparation using [3H]saxitoxin binding assays, and the density of acetylcholine receptors determined in fractions from control and denervated muscle using alpha-[125I]bungarotoxin. Although both fractions contain predominantly surface membrane, the LiBr fraction consistently shows the higher specific activity of p-nitrophenylphosphatase, higher free cholesterol content, and higher density of sodium channels and acetylcholine receptors. The density distribution of sodium channels appears uniform throughout both fractions. Quantitative differences were seen between sodium dodecyl sulfate polyacrylamide gel electrophoresis patterns of membrane proteins from the two preparations although most bands are represented in both. A majority of the low-salt sucrose light membrane proteins were accessible in varying degrees to labelling with diazotized diiodosulfanylic acid in intact muscle. These results suggest that light surface membrane fractions may be mixtures of sarcolemma and T-tubular membranes. Using our preparative methods, the LiBr fraction may contain predominantly sarcolemma while low-salt sucrose light membranes may be enriched in T-tubular elements.