Recent effort to build an unmanned pollen monitoring network in Japan has led to new developments in automatic pollen counters. In-the-field performance tests of these automatic counters have not been reported. To characterize recently developed automatic pollen counters, with a view of using their data in pollen information systems. We performed side-by-side comparisons between 2 recently developed automatic pollen counters and 2 reference samplers at 2 sites during the 2005 pollen season. Both automatic counters were found to have similar overall performance in terms of their correlations with the reference samplers. The linear correlation coefficient between the hourly values of the counters and one of the reference samplers was larger than 0.8 at both sites for both counters. Although these results are encouraging, our analysis also points to weaknesses of the investigated automatic counters in the areas of pollen discrimination, minimum measurable concentration, and calibration. Both counters were found to be affected by large concentrations of particulate matter, although the conditions and extent to which the particulate matter disrupted the measurements differ for the 2 sensors. The effect of particulate matter is particularly noticeable at the start and end of the pollen season, that is, when pollen concentration is low relative to particulate matter concentration. Further, it was found that one of the automatic counters could not differentiate snow particles from pollen grains. The tested automatic pollen counters had good overall performances, but weaknesses in the areas of pollen discrimination, minimum measurable concentration, and calibration still have to be addressed for these counters to find widespread use in the allergy community.