Palynological assemblages from two outcrops of the upper part of the Memorial Formation, the Lost Branch Formation, and the overlying Hepler unit in Kansas were examined to discover which stratigraphic interval marks the change from the lycopod-dominated coal swamp floras of Middle Pennsylvanian (Westphalian D) age to the fern-dominated coal swamp floras of Late Pennsylvanian (Stephanian) age. The Lost Branch Formation underlies the Pleasanton Group, whose base is recognized as the Middle-Upper Pennsylvanian boundary in the Midcontinent. The outcrops include the youngest Middle Pennsylvanian coal (Dawson), just below the Lost Branch Formation, and the oldest Upper Pennsylvanian coal ('Hepler') within the Pleasanton Group. Lycospora dominates the spore assemblage in the Middle Pennsylvanian (Desmoinesian) Dawson coal in the Memorial Shale and is abundant in shale between the coal and just below the Glenpool limestone bed at the top of the Lost Branch Formation. It is rare between the limestone and the Upper Pennsylvanian (Missourian) 'Hepler' coal. Granasporites medius and Thymospora pseudothiessenii disappear below the limestone. The 'Hepler' coal is dominated by fern and seed fern spores Cyclogranisporites and Apiculatasporites, and the sphenopsid spore Calamospora is third in abundance. Florinites, Potonieisporites and other gymnospermic monosaccate pollen are abundant between the two coals. Bisaccate conifer-like pollen, such as Protohaploxipinus, are most common between the Dawson coal and Glenpool limestone, but Wilsonites, which is thought to have been produced by seed ferns, is very abundant from the Glenpool limestone to the 'Hepler' coal. On the basis of macroinvertebrate evidence, the Glenpool limestone is Middle Pennsylvanian in age, but the palynological evidence indicates that the floral change took place slightly before deposition of the limestone. Thus, the major change in climate that occurred near the Middle-Upper Pennsylvanian boundary apparently affected the floras earlier than the faunas. The floral change cannot be explained as resulting from a major marine regression and hiatus because the change is recorded in a marine section within a single transgressive-regressive sequence.