Isolated equid phalanges are relatively common finds in the Barstow Formation (Miocene, 19 to 13 million years ago, southern California), but anecdotal observations suggested that not all positions (proximal, middle, and distal/ungual) of the primary digit (digit III) are recovered with equal frequency. Our sample includes primarily surface-collected phalanges from the Barstow Formation, which we compare with phalanges of Pleistocene horses from Pit 3 and Pit 77 from Rancho La Brea (Los Angeles, California). The null hypothesis is that the three positions of phalanges should be equally common. Our Barstow sample includes 228 proximal, 151 middle, and 36 distal phalanges. A chi-square test (p<0.001) is consistent with preservation bias in phalangeal frequency for the full Barstow Formation sample, and this pattern generally holds within sub-samples by locality or depositional environment. Pit 3 of La Brea produced 163 proximal, 144 middle, and 103 distal phalanges. A chi-square test with correction for multiple comparisons suggests that proximal and middle phalanges are preserved with statistically equal frequency (p=0.278) whereas distal phalanges are less common (p<0.001). For Pit 77, there are 54 proximal, 55 middle, and 51 distal phalanges; the chi-square test finds that all three types are equally common (p=0.922). Overall, differences in physical properties between phalangeal positions, such as surface area, density, shape, and size, could influence preservation within each environment. The observed differences between Barstow and La Brea might be caused by variations in depositional environment that influence the surface exposure time of fossils and disarticulation pre-burial, as well as by differences in the size of the horses at each locality. We suggest that when permitted by sample size, it is desirable to distinguish unguals from other phalanges when analyzing taphonomic patterns in the fossil record.