BACKGROUND Diatoms belong to the stramenopiles, one of the largest groups of eukaryotes, which are primarily characterized by a presence of an anterior flagellum with tubular mastigonemes and usually a second, smooth flagellum. Based on cell wall morphology, diatoms have historically been divided into centrics and pennates, of which only the former have flagella and only on the sperm. Molecular phylogenies show the pennates to have evolved from among the centrics. However, the timing of flagellum loss--whether before the evolution of the pennate lineage or after--is unknown, because sexual reproduction has been so little studied in the 'araphid' basal pennate lineages, to which Pseudostaurosira belongs. RESULTS Sexual reproduction of an araphid pennate, Pseudostaurosira trainorii, was studied with light microscopy (including time lapse observations and immunofluorescence staining observed under confocal scanning laser microscopy) and SEM. We show that the species produces motile male gametes. Motility is mostly associated with the extrusion and retrieval of microtubule-based 'threads', which are structures hitherto unknown in stramenopiles, their number varying from one to three per cell. We also report experimental evidence for sex pheromones that reciprocally stimulate sexualization of compatible clones and orientate motility of the male gametes after an initial 'random walk'. CONCLUSIONS The threads superficially resemble flagella, in that both are produced by male gametes and contain microtubules. However, one striking difference is that threads cannot beat or undulate and have no motility of their own, and they do not bear mastigonemes. Threads are sticky and catch and draw objects, including eggs. The motility conferred by the threads is probably crucial for sexual reproduction of P. trainorii, because this diatom is non-motile in its vegetative stage but obligately outbreeding. Our pheromone experiments are the first studies in which gametogenesis has been induced in diatoms by cell-free exudates, opening new possibilities for molecular 'dissection' of sexualization.