We present new major and trace element data for a set of closely spaced (< 1.8 km) dredges along a well‐studied portion of the East Pacific Rise (EPR) axis near 90°30′N (9°17′N to 90°50′N). With the exception of enriched mid‐ocean ridge basalt (E‐MORB) at 9°35′N, the lavas are all normal mid‐ocean ridge basalt (N‐MORB) with a limited range of MgO (8.40–6.22 wt %). Major element and trace element data favor derivation of the melts from a single parental composition by low pressure crystallization of olivine, plagioclase and clinopyroxene in the ratio of 16:62:22. This model is consistent with liquid line of descent models but inconsistent with petrographic observations in that most of the N‐MORB lavas have only plagioclase phenocrysts. We ascribe this to gravitational crystal settling of mafic phases and flotation of plagioclase, supported by crystal size distribution data and density relations. Most likely this occurred in the axial magma chamber (AMC) that underlies the EPR in this area. The chemistry of axial lavas varies along axis and correlates roughly with elevation of the axis and depth to the AMC. We interpret these correlations as favoring an AMC that is chemically zoned along‐axis, with Fe‐rich melts at its distal ends. This favors a central injection of MgO‐rich melt with lateral along‐axis shallow transport. The height of eruptions along axis is apparently controlled by magma density such that least dense MgO‐rich melts build local volcanic constructs of the highest elevation. The E‐MORB lavas are older than the N‐MORB and probably erupted at a time when the AMC was absent or was much smaller in size than presently. E‐MORB could have originated by deep melting processes or very shallow contamination of N‐MORB. Its presence in the 9°30′N area supports the notion that magma chambers are not truly steady state. Instead they probably come and go on a time scale of 3000–6000 years. A single parental magma seems to supply melts to the AMC along the entire 60 km segment of the EPR, suggesting that central supply injection sites are widely spaced. Based on our data, there is no evidence for petrologic segmentation corresponding to 4th‐order tectonic segmentation in this area; however, it may be present below the resolution of our sampling.