Abstract
The use of a single loop-detector system on a multi-lane road is bound to result in undercounting. This report describes a theoretical method of calculating the probability of a vehicle not being detected under these circumstances and compares the results with 'real' traffic count data collected by several authorities. For all the data examined the theoretical approach gives similar undercounting errors to those actually observed and hence is a useful method of estimating these errors when 'real' data is not available. (Author/publisher)