A spatio-temporal near-neighbor model is used to identify and map variations in the recurrence rate of volcanism in the Springerville volcanic field, Arizona, a large field on the Colorado Plateau boundary. Detailed mapping of individual lava flows and their associated vents, together with radiometric and paleomagnetic dating, demonstrates that 366 volcanic events have formed the Springerville volcanic field. These volcanic events consist of mapped units that erupted between 2.1 and 0.3 Ma over an area of 3000 km2. Cumulatively, the rate of vent formation in the Springerville field waxed prior to 1.5 Ma, was near steady state from 1.5 to 0.75 Ma, and has waned since 0.75 Ma. The increase in the rate of vent formation at ca. 1.5 Ma coincided with a shift in the locus of Springerville magmatism from west to east and an increase in the alkalic nature of the magma, including eruption of mugearites and benmoreites. The volume of flows, inferred from lava-flow areas, was steady state from 1.75 to 0.75 Ma. A near-neighbor spatio-temporal recurrence-rate model using seven near-neighbor volcanoes and a 0.5 m.y. time window reveals that (1) areas of waxing and waning magmatism in the Springerville volcanic field are much more localized and (2) volcanic activity within these areas is much more intense than implied by field-wide temporal trends. These volcano clusters are 10–20 km in diameter; they were commonly active for less than 0.25 m.y. Mugearites and benmoreites are limited to these areas of high recurrence rate. This clustered and petrologically distinctive, rather than distributed or random, volcanic activity suggests that individual source regions for the magma also are localized and short-lived compared with the area and longevity of the entire field. Because volcanic activity is spatially and temporally clustered, forecasting subsequent activity is more successful if the spatio-temporal recurrence-rate model is used, rather than the average recurrence rates. This success indicates that spatio-temporal recurrence-rate models are useful tools for the quantification of long-term volcanic hazards in basaltic volcanic fields.