A historical prospective mortality study was conducted at a glass filament plant in Ontario, Canada. The cohort consisted of 1465 men and women who had worked for a total of at least one year between 1951 (when the operations began) and 1986. Ninety six deaths were found in the 96% of persons traced. Mortality was compared by the person-years method with that of the Ontario population. An estimate of cumulative dust exposure was made based on long term employees' recollections of past conditions. Overall mortality was below that expected (standardised mortality ratio (SMR) = 76) based on 82 deaths. There were 11 deaths due to lung cancer; slightly but not significantly more than expected (SMR = 136, p = 0.31). Analysis by both duration of employment and cumulative exposure showed this increase to be inconsistent with an occupational aetiology.