To easily automate the abnormality diagnosis of water quality instruments by obtaining the measured values of plural water quality instruments of different types at each prescribed measuring cycle.

Measured values Xk (k=1-n) of (n) water quality instruments different in type at each prescribed measuring cycle are obtained, and the absolute values ΔXk (k=1-n) of the deviations between the measured values Xk and the previous measured values Xk' are obtained for (n) water quality instruments. ΔXk=0 when ΔXk≤Ek, and ΔXk=1 when ΔXk>Ek, and the values A and B indicated by the equations I and II are obtained, where Ek (k=1-n) are the assumed measurement errors for the (n) water quality instruments. Whether A=0 or not and whether B=0 or not are judged. If A≠0 and B=0, it is judged that there is an abnormality. If A≠0 and B≠0, it is judged that the water quality is changed.