Description: Introduction Since 2000 a decline in the incidence of occupational asthma (OA) has been reported in the UK and Europe. We aimed to describe and account for trends in the incidence of OA in the West Midlands, UK using annual notification data from the SHIELD voluntary surveillance scheme over the period 1991–2011. Methods All notifications to the SHIELD database between January 1991 and December 2011 were identified, along with patients’ demographic data, occupations, causative agents and confirmatory tests. Annual notifications were scaled to give an annual count per million workers, giving a measure of incidence, and also standardised against those of bakers’ asthma. Non-parametric analyses were undertaken between annual incidence and time (years) for common causative agents using (1) a negative binomial regression univariate model and (2) a logistic regression model calculating annual reporting ORs. A step-change analysis was used to examine time points at which there were marked reductions in incidence. Results A decrease in annual incidence of OA was observed over the study period (incident rate ratio=0.945; 95% CI 0.933 to 0.957; p〈0.0001), an effect that was lost after standardising for bakers’ asthma. Decreases in incidence were seen for most common causative agents, with only cleaning product-related OA increasing over 21 years. Marked fall in incidence was seen in 2004 for isocyanates, and in 1995 for latex. Most notifications came from a regional specialist occupational lung disease unit, with notifications from other sites falling from 16 cases/million workers/annum in 1995 to 0 in 2004. Conclusions Reporter fatigue and increasing under-recognition of OA are both factors which contribute to the apparent fall in incidence of OA in the West Midlands. There is a future need for interventions that enable health professionals to identify potential cases of OA in the workplace and in healthcare settings.