Fire blight of pear and apple is frequently an inoculum-limited disease but weather-based forecasting models commonly assume that the pathogen is omnipresent. To improve fire blight risk assessment during flowering, we developed a rapid pathogen detection protocol that uses loop-mediated isothermal amplification (LAMP) to detect DNA of epiphytic Erwinia amylovora on samples of pear and apple flowers. LAMP detected a single flower colonized epiphytically by E. amylovora in a sample of 100 flower clusters (approximately 600 flowers). Samples of 100 flower clusters from orchards (approximately one sample per hectare) were washed and subjected to LAMP, which was completed in 2 h. In three experimental orchards inoculated with E. amylovora, positive LAMP reactions were attained from nine of nine 100-flower cluster samples; pathogen populations in the floral washes averaged 5.2 × 103 CFU per flower as determined by dilution plating. Samples of pear and apple flowers obtained from 60 commercial orchards located in Oregon, Washington, California, and Utah resulted in detection of E. amylovora by LAMP assay from 34 sites, 20 of which developed fire blight. Of samples at early bloom, 10% were positive for epiphytic E. amylovora compared with 28% at petal fall; pathogen density in washes of positive samples averaged 3.2 × 102 CFU per flower. In another 26 orchards, all floral washes were negative for E. amylovora by LAMP and by dilution plating; a light severity of fire blight was observed in 8 of these orchards. Overall, positive detection of epiphytic E. amylovora in commercial orchards by LAMP-based scouting generally occurred at later stages of bloom after heat (risk) units had begun to accumulate, an indication that weather-based forecasting models may be an adequate measure of fire blight risk for many orchardists. Nonetheless, several orchardists communicated that information from the LAMP-based rapid detection protocol resulted in modification of their fire blight management practices.