Hourly values of relative humidity recorded at 75 stations across Canada were examined. Data were checked for possible discontinuities due to changes in procedures and instruments. Notable changes included a refinement of the psychrometric tables around 1960 and the replacement of the psychrometer by the dewcel in the early 1970s. The latter is shown to have caused a decreasing step in relative humidity at a number of stations. The statistical significance of the step was determined by comparing two regression models using the F-test. This step particularly affects the relative humidity trends for stations experiencing cold temperatures. The impact of the change in instruments on seasonal trends during the period 1953-2003 will be presented.