Models of urban spatial structure take on a life of their own. We still refer to the Concentric Zone Model of Burgess and Hoyt even though it was developed to describe Chicago during the mid-1920s. This is not totally bad. Good models, those that illustrate the processes of urban growth and change in straightforward form, are flexible and can be modified to fit particular settings. As heuristic devices they teach conceptual frameworks at very basic levels. The model of Latin American city structure by Ernst Griffin and myself that was published in the Geographical Review in 1980 has held up well, given the number of times it has appeared in citations. Comprehensible at a glance, the model can be adjusted to fit particular local conditions. On the other hand, cities do change, especially in rapidly growing regions such as Latin America, so it stands to reason that models ought to change as well. In addition, models profit from modification every now and then, if only to include some of the insights and suggestions tendered by authors who have critiqued them over time. I therefore offer a new and improved model of Latin American city structure, one which - I hope - combines the simplicity and process orientation of the original with a few important new urban characteristics [ILLUSTRATION FOR FIGURE 1 OMITTED]. Most of the other models of Latin American city structure that have been put forward are more complex and elaborate than the one we designed. For example, the German geographers Jurgen Bahr and Gunter Mertins included not only a far larger assortment of morphological districts in their models but also arrows aimed at adding a dynamic element (Bahr 1976; Bahr and Mertins 1981). Though insightful, these models tend to be unnecessarily complicated, with major patterns and processes given the same attention as minor intracity flows. The model of Mexican border cities put forward by Daniel D. Arreola and James R. Curtis (1993) provides another example of complexity, with three times as many districts and features as our 1980 version. Atop the complexity heap is William K. Crowley's (1995) attempt to create a model based on the merger of three separate submodels portraying commercial, industrial, and residential land uses. The result is a complete depiction of each possible element of Latin American city structure at an overwhelming level of detail - rather as if a map of the London subway system had been printed on an Indonesian batik fabric. Although these and other models are too complex, it is also possible that our 1980 version is now too simple. The trick is to add new key elements without destroying heuristic simplicity and process orientation. A model, after all, is not a map: Not everything need be shown. With that in mind, I offer the modified model depicted in Figure 1. The new model retains the simple, circular form of the original and its essential elements - central business district (CBD), commercial spine, elite residential sector, three concentric rings of diminishing residential status (Zone of Maturity, Zone of In Situ Accretion, and Zone of Peripheral Squatter Settlements), and sectors of disamenity. Though not cartographically depicted in the model, all of the zones are assumed to contain mixed uses to some extent. For example, all residential areas include retail, eating, and recreational establishments, and many contain small industrial concerns. These cannot be included visually without creating a mess. On the other hand, some changes can be made without adding too much complexity. The new model incorporates six changes, as follows: * The new model's downtown is divided into two parts: CBD and Market. This reflects the fact that many Latin American downtowns now have a number of modern office, hotel, and retail structures quite separate from the more traditional and mixed market districts. The contrast between small, street-oriented businesses and self-contained megastructures suggests an increasing split of the downtown into modern and traditional sections. …