Measured soil moisture data may improve wildfire probability assessments because soil moisture is physically linked to fuel production and live fuel moisture, yet models characterising soil moisture-wildfire relationships have not been developed. We therefore described the relationships between measured soil moisture (concurrent and antecedent), as fraction of available water capacity (FAW), and large (≥405 ha) wildfire occurrence during the growing (May-October) and dormant (November-April) seasons from 2000 to 2012 in Oklahoma, USA. Wildfires were predominantly grass and brush fires but occurred across multiple fuel types including forests. Below-average FAW coincided with high wildfire occurrence each season. Wildfire probability during the growing season was 0.18 when concurrent FAW was 0.5 (a threshold for plant water stress) but was 0.60 when concurrent FAW was 0.2 (extreme drought). Dormant season wildfire probability was influenced not only by concurrent but also by antecedent FAW. Dormant season wildfire probability was 0.29 and 0.09 when FAW during the previous growing season was 0.9 (near ideal for plant growth) and 0.2, respectively. Therefore, although a wet growing season coincided with reduced wildfire probability that season, it also coincided with increased wildfire probability the following dormant season, suggesting that the mechanisms by which soil moisture influences wildfire probability are seasonally dependent.