Assessments of direct smears of synovial fluid by four clinicians were comp
ared with the results obtained with a Coulter counter. Estimates of total w
hite cell counts by the clinicians were inaccurate and generally higher tha
n the Coulter counter results. The method had a low sensitivity and specifi
city for the identification of degenerative joint disease and normal joints
in comparison with the identification of inflammatory joint disease. There
were marked variations in the results obtained by the four clinicians for
all the analyses in the study.