The object of the invention is to provide a method for counting the number of nucleic acid molecules in a mixture of a plurality of nucleic acid molecules more highly accurately. This is a method for highly accurately counting the number of nucleic acid molecules by detecting the read errors that occur when determining a nucleic acid base sequence, wherein the method has: a step for adding a barcode-sequence-generating oligonucleotide to a mixture of a plurality of nucleic acid molecules, thereby linking barcode sequences unique to the nucleic acid molecules to the base sequences constituting each of the nucleic acid molecules; a step for determining the base sequences of the nucleic acid molecules to which the barcode sequences have been linked; a step for detecting read errors in the barcode sequences for which the base sequences have been determined; and a step for calculating the proportion of barcode sequences free of read errors to all of the barcode sequences for which the base sequences have been determined, on the basis of the number of reads of the barcode sequences for which the base sequences have been determined, the abovementioned barcode-sequence-generating oligonucleotide comprising a maximum of five bases, and the number of barcode sequences free of read errors indicating the number of nucleic acid molecules in the mixture.