Between 1876 and 1913, diamonds were found in at least seven localities in southern and central Wisconsin. All were found in Pleistocene glacial deposits or Holocene river gravel. The bedrock kimberlite source for the diamonds is unknown but has been presumed to be in northern Canada, the only area north of Wisconsin previously known to contain kimberlites. Recently, a kimberlite pipe, here named the Lake Ellen kimberlite, has been found in Iron County, Michigan. That find suggests the possibility that drift diamonds in Wisconsin have come from a more local source--kimberlites in northern Michigan and Wisconsin. The Lake Ellen kimberlite is very poorly exposed, but a strong positive magnetic anomaly indicates that it is roughly circular in plan and about 200 m in diameter. Although the kimberlite is entirely surrounded by Precambrian rocks, it contains abundant inclusions of fossiliferous dolomite, probably from the Ordovician Black River Group that overlay the area when the kimberlite was intruded. The post-Ordovician age of the kimberlite leads us to suspect that other possible cryptovolcanic structures in Paleozoic rocks in the region were formed over kimberlite pipes that are not yet exposed by erosion. Such structures include Limestone Mountain and Sherman Hill, in Houghton and Baraga Counties, Michigan; Glover Bluff, in Marquette County, Wisconsin; and possibly an area along the Brule River south of Iron River, Michigan. No diamonds are known in the Lake Ellen kimberlite, but it has not been adequately sampled. The cryptovolcanic structures could not be the source of the drift diamonds in Wisconsin because even if the structures are caused by kimberlites, those kimberlites have not yet been exposed by erosion. Elsewhere in the world, kimberlite is seldom found as a single isolated body; clusters of bodies are more common, and the presence of one kimberlite implies that others may exist nearby. The discovery of additional kimberlites may be very difficult because of the extensive cover of glacial drift and the typical small size of kimberlite bodies. If all are magnetic, they might be found by detailed aeromagnetic surveys. However, the magnetism of the Lake Ellen kimberlite appears to be caused by secondary magnetite formed during serpentinization of olivine, so an unserpentinized kimberlite may not be strongly magnetic. We suggest that one or more diamond-bearing kimberlites may exist in northern Michigan or Wisconsin, but the discovery of such bodies is unlikely unless a very thorough search is undertaken.