In 2016, Kaji et al. concluded that the amphioxus mouth has the quality of a coelomoduct and is, therefore, not homologous to the oral opening of any other animal. They studied a Japanese population of Branchiostoma japonicum and based their conclusion, in part, on the larval expression of BMP2/4 in cells that reportedly joined the rim of the forming mouth. They did not detect transcription of that gene in any other tissues in the anterior region of the larva. Their results were almost the inverse of findings for B. floridae by Panopoulou et al. (1998), who detected BMP2/4 expression in several anterior tissues, but not in cells intimately associated with the nascent mouth. To resolve this discrepancy, we have studied BMP2/4 in a Chinese population of B. japonicum as well as in an additional species, the European B. lanceolatum. In both species, larval expression of BMP2/4 closely resembles the pattern previously reported for B. floridaedthat is, transcription is undetectable in tissues juxtaposed to the forming mouth, but is seen in several other anterior structures (most conspicuously in the lining of the rostral coelom and the club-shaped gland). In sum, we could not repeat the BMP2/4 expression pattern of Kaji et al. (2016) even in the same species, and their findings for this gene, at least, cannot be counted as a support for their hypothesis for a coelomoduct mouth.