The localisation of maize (Zea mays L.) auxin-binding protein (ABP1) has been studied using a variety of techniques. At the whole-tissue level, tissue printing indicated that ABP1 is expressed to similar levels in all cells of the maize coleoptile and in the enclosed leaf roll. Within cells, the signals from immunofluorescence and immunogold labelling of ultrathin sections both indicated that ABP1 is confined to the endoplasmic reticulum (ER), none being detected in either Golgi apparatus or cell wall. This distribution is consistent with targeting motifs in its sequence. These observations are discussed with reference to the various reports which place a population of ABP1 on the outer face of the plasma membrane, including those suggesting that it is necessary on the cell surface for rapid, auxin-mediated protoplast hyperpolarisation. We have tested the ER, namely that auxin binding induces a conformational change in ABP1 leading to concealment of the KDEL retention motif. Using double-label immunofluorescence the characteristic auxin-induced rise in Golgi-apparatus signal was found, yet no change in the distribution of the ABP1 signal was detected. Maize suspension cultures were used to assay for auxin-promoted secretions of ABP1 into the medium, but secretion was below the limit of detection. This can be ascribed at least partly to the very active acidification of the medium by these cells and the instability of ABP 1 in solution below pH 5.0. In the insect-baculovirus expression system, in which cell cultures maintain pH 6.2, a small amount of ABP1 secretion, less than 1% of the total, was detected under all conditions, Insect cells were shown to take up auxin and no inactivation of added auxin was detected, but auxin did not affect the level of ABP1 in the medium. Consequently, no evidence was found to support the model for auxin promotion of ABP1 secretion. Finally, quantitative glycan analysis was used to determine what proportion of ABP1 might reach the plasma membrane in maize coleoptile tissue. The results suggest that less than 15% of ABP1 ever escapes from the ER as far as the cis-Golgi and less than 2% passes further through the secretory pathway. Such leakage rates probably do not require a specialised mechanism allowing ABP1 past the KDEL retrieval pathway, but we are not able to rule out the possibility that some ABP1 is carried through associated with other proteins. The data are consistent with the presence of ABP1 both on the plasma membrane and in the ER. The relative sizes of the two pools explain the results obtained with immunofluorescence and immunogold labelling and illustrate the high efficiency of ER retention in plants.