A mathematically determined technique which seemed to be valid for the calculation of crowding and spacing in arches with acceptable alignment, was tested on a series of arches with varying degrees of crowding and irregularity. Upper and lower study casts of 30 individuals with sufficient malalignment and crowding to warrant orthodontic treatment with fixed appliances were examined. Models were measured by three operators, using a reflex microscope interfaced to an IBM compatible PC and a customized calculation program. The degree ofcrowdmg was compared with a ‘clinical’ assessment of each arch, in which the irregularity was measured directly using a steel ruler. Both the computer calculated and clinical values varied between observers, but the agreement between observers was better using the reflex microscope. Correlation coefficients ranged from 0·92 to 0·96 for the reflex microscope, and between 0·86 and 0·91 for the clinical assessment. Inter-observer agreement for the three operators was high: 0·93 for the reflex microscope and 0·87 where a ruler was used. This suggests that the computer program used to calculate the crowding was valid for the crowded arches under consideration.