Ploidy levels of in vitro shoots produced through anther culture of asparagus were determined by flow cytometry (FCM) analysis instead of counting the chromosome number of root tips that were difficult to regenerate. Of the 110 anther-derived shoots, there were 83 diploids, 21 tetraploids, 3 octaploids, and 3 mixoploids, but no haploids. When flower buds of different length were used as starting material for anther culture, the richest variation in polyploidy originated from short flower buds (1.0-1.5 mm). This result suggested that polyploids might develop by the fusion of nuclei during pollen-mother cell division before the tetrad stage. We used scanning electron microscopy (SEM) to measure the stomatal length of in vitro and field-grown shoots in order to clarify the differences among polyploids determined by FCM analysis. When the developmental stage of tissue was taken into account, we found statistically significant differences in stomatal length among all polyploids (diploid, tetraploid, and octaploid) for both stems and phyllodes (P < 0.001). Stomata on stems were longer than those on phyllodes, and mature tissue had longer stomata than immature tissue. Morphological observations with SEM found that immature tissue had guard cells set below epidermal cells, and were poorly developed. Field-grown diploids had much shorter stomatal length than in vitro diploids. Male plants grown in the field had significantly longer stomata than female plants in both their immature stems and mature phyllodes (P < 0.001). These results indicated that stomatal length can be a reliable ploidy index by removing any other influential factors such as sex, environment and developmental stage.