The smectite-to-chlorite conversion is investigated through long-duration experiments (up to 9 years) conducted at 300 °C. The starting products were the Wyoming bentonite MX80 (79 % smectite), metallic iron and magnetite in contact with a Na–Ca chloride solution. The predominant minerals in the run products were an iron-rich chlorite (chamosite like) and interstratified clays interpreted to be chlorite/smectite and/or corrensite/smectite, accompanied by euhedral crystals of quartz, albite and zeolite. The formation of pure corrensite was not observed in the long-duration experiments. The conversion of smectite into chlorite over time appears to take place in several steps and through several successive mechanisms: a solid-state transformation, significant dissolution of the smectite and direct precipitation from the solution, which is over-saturated with respect to chlorite, allowing the formation of a chamosite-like mineral. The reaction mechanisms are confirmed by X-ray patterns and data obtained on the experimental solutions (pH, contents of Si, Mg, Na and Ca). Because of the availability of some nutrients in the solution, total dissolution of the starting smectite does not lead to 100 % crystallization of chlorite but to a mixture of two dominant clays: chamosite and interstratified chlorite/smectite and/or corrensite/smectite poor in smectite. The role of Fe/(Fe + Mg) in the experimental medium is highlighted by chemical data obtained on newly formed clay particles alongside previously published data. The newly formed iron-rich chlorite has the same composition as that predicted by the geothermometer for diagenetic to low-grade metamorphic conditions, and the quartz + Fe-chlorite + albite experimental assemblage in the 9-year experiment is close to that fixed by water–rock equilibrium.