We examine the occurrences, textures, and compositional patterns of spinels in the olivine- phyric shergottites Sayh al Uhaymir (SaU) 005, lithology A of Elephant Moraine A79001 (EET-A), Dhofar 019, and Northwest Africa (NWA) 1110, as well as the lherzolitic shergottite Allan Hills (ALH) A77005, in order to identify spinel-olivine-pyroxene assemblages for the determination of oxygen fugacity (using the oxybarometer of Wood ) at several stages of crystallization. In all of these basaltic martian rocks, chromite was the earliest phase and crystallized along a trend of strict Cr-Al variation. Spinel (chromite) crystallization was terminated by the appearance of pyroxene but resumed later with the appearance of ulvospinel. Ulvospinel formed overgrowths on early chromites (except those shielded as inclusions in olivine or pyroxene), retaining the evidence of the spinel stability gap in the form of a sharp core/rim boundary (except in ALH A77005, where subsolidus reequilibration diffused this boundary). Secondary effects seen in chromites include reaction with melt before ulvospinel overgrowth, reaction with melt inclusions, reaction with olivine hosts (in ALH A77005), and exsolution of ulvospinel or ilmenite. All chromites experienced subsolidus Fe/Mg reequilibration. Spinel-olivine-pyroxene assemblages representing the earliest stages of crystallization in each rock essentially consist of the highest-Cr#, lowest-fe# chromites not showing secondary effects plus the most magnesian olivine and equilibrium low-Ca pyroxene. Assemblages representing the onset of ulvospinel crystallization consist of the lowest-Ti ulvospinel, the most magnesian olivine in which ulvospinel occurs as inclusions, and equilibrium low-Ca pyroxene. The results show that, for early crystallization conditions, oxygen fugacity (fO2) increases from SaU 005 and Dhofar 019 (~QFM -3.8), to EET-A (QFM -2.8) and ALH A77005 (QFM -2.6), to NWA 1110 (QFM -1.7). Estimates for later conditions indicate that in SaU 005 and Dhofar 019 oxidation state did not change during crystallization. In EET-A, there was an increase in fO2 that may have been due to mixing of reduced material with a more oxidized magma. In NWA 1110, there was a dramatic increase, indicating a non-buffered system, possibly related to its high oxidation state. Differences in fO2 among shergottites are not primarily due to igneous fractionation but, rather, to derivation from (and possibly mixing of) different reservoirs.