NanoSIMS measurements of Ti, P, Y, Ce, and Hf in zircon separates from the Youngest Toba Tuff, the Bishop Tuff, the Quottoon Igneous Complex, the Sierra Nevada batholith, and an Adirondack migmatite show that micron-scale oscillatory zoning of Ti is common. The zircons we have studied typically display banded concentration gradients, having between 1 and 7 peaks in Ti abundance over length scales of 10–20 μm—the beam diameter commonly used for SIMS trace-element zircon analyses—with amplitudes of up to 4.5 ppm Ti (baseline values are ~1–4 ppm) and widths (measured at half-height) of between ~0.2 and 4 μm. Spatial correlations between concentrations of Ti and other trace elements (P, Y, and Ce) are also common, but variable in character, ranging from oscillatory co-variation of Ti, P, Y, and Ce to cases where only a subset of peaks for a given element is spatially correlated with peaks in Ti. There are also longer length-scale, generally positive correlations among concentrations of Ti, P, Y, and Ce (i.e., gradients on which narrower peaks are superimposed). In contrast, Hf concentrations are either uncorrelated or inversely correlated with these longer length-scale variations in Ti concentrations. The wide range in Ti concentrations over distances of less than 1 μm and the various correlations between Ti concentrations and those of the other analyzed elements suggest that on the micron scale and at temperatures between ~700 and 800 °C, zircon-liquid Ti partitioning is not controlled by bulk or lattice equilibrium. Treating our NanoSIMS Ti ion maps as though they were conventional SIMS analyses (i.e., generating an average Ti concentration for each map), we evaluate the hypothesis that while micron-scale variations in Ti concentrations might be kinetically controlled, when averaged over 100–400 μm 2 , such variations capture the thermal state of the growing zircon. Using these average Ti concentrations, independent petrologic pressure and temperature constraints, estimates of silica and titania activities based on phase assemblages as well as calculations using rhyolite-MELTS, we show that crystallization temperatures predicted by Ti-in-zircon geothermometry generally do not agree with the independently constrained temperatures for the samples.