A new approach for process monitoring is described, the self-organizing map quality control (SOMQC) index. The basis of the method is that SOM maps are formed from normal operating condition (NOC) samples, using a leave-one-out approach. The distances (or dissimilarities) of the left out sample can be determined to all the units in the map, and the nth percentile measured distance of the left out sample is used to provide a null distribution of NOC distances which is generated using the Hodges-Lehmann method. The nth percentile distance of a test sample to a map generated from all NOC samples can be measured and compared to the null distribution at a given confidence level to determine whether the sample can be judged as out of control. The approach described in this paper is applied to online high-performance liquid chromatography (HPLC) measurements of a continuous pharmaceutical process and is compared to other established methods including Q and D statistics and support vector domain description. The SOMQC has advantages in that there is no requirement for multinormality in the NOC samples, or for linear models, or to perform principal components analysis (PCA) prior to the analysis with concomitant issues about choosing the number of PCs. It also provides information about which variables are important using component planes. The influence of extreme values in the background data set can also be tuned by choosing the distance percentile.