The homogeneity of total solar radiation during 1959-2012 in Tianjin is examined using the penalized maximal t-test and a sequential algorithm, using data and metadata of total solar radiation and the percentage of sunshine on monthly and annual scales. Results indicate that instrument change is mainly responsible for three shifts in total solar radiation at or around 1968, 1972, and 1990, while station relocation does not play a significant role. Moreover, a gradual decline in total solar radiation in the early 1980s, around 1983, is detected using the sequential algorithm. The non-homogeneity effect of instrument change is also detected in monthly total solar radiation series. Therefore, in studies of solar radiation, we must not only respect the raw data, but also fully consider the homogeneity of the data. Excluding non-homogeneity is necessary to obtain reliable analyses of climate change.