Paleomagnetic inclinations of lavas and dikes in CY-1 and CY-1A (Troodos ophiolite) and 504B (Costa Rica Rift) drill cores show remarkably similar patterns of postdepositional rotation about a horizontal axis. We interpret these patterns in terms of architecture of upper ophiolite and oceanic crust, lava deposition, and deformation at the paleo-spreading center. We recognize two components of postdepositional rotation about an axis parallel to the spreading axis: a variable rotation with depth due to lava-flow loading (zero at the top of the lava pile and generally increasing with depth) and a depth-invariant block rotation. At both drill sites, maximum rotation due to lava-flow loading dominates over estimated block rotation. Rotation (toward the spreading axis) due to lava-flow loading increases from zero at the top of the lava pile to a maximum of approximately 50 degrees -70 degrees at the base. Block rotation (away from the axis of spreading) is approximately 0 degrees in 504B and approximately 20 degrees in CY-1 and CY-1A. The component due to lava-flow loading predicts a narrower zone of lava deposition at the Costa Rica paleoridge ( approximately 3 km) than at the Troodos paleoridge ( approximately 5 km). This difference suggests that the Troodos spreading center had a wider rift valley or was more like the East Pacific Rise where lavas can flow several kilometers down the flanks unobstructed by large inward-facing fault scarps. The fact that the sheeted dikes in Hole 504B and in the surface geology down section from Holes CY-1 and CY-1A are unrotated relative to the top of the lava pile implies an abrupt change in dip of 50 degrees -70 degrees between the sheeted dikes and the dikes in the transition zone from sheeted dikes to overlying lava flow dikes (where most dikes are orthogonal to the lava flows). We suggest that this angular discontinuity near the base of the transition zone at both sites represents a horizontal decoupling horizon that partitions deformation between the lavas and sheeted dikes. This interpretation is an alternative to the traditional views that the discontinuity is evidence of a normal fault that cuts 504B and the axis of an abandoned spreading center that crosses the Akaki River section of the Troodos ophiolite at this stratigraphic level. We speculate that the contrasting kinematic histories recorded in the lava section and the sheeted dikes in Hole 504B and in and near Holes CY-1 and CY-1A reflect their contrasting mechanical response to subsidence due to lava-flow loading, i.e., the lava pile deformed primarily by bending and the sheeted dike section deformed by vertical slip along closely spaced planes parallel to the dikes.