Better information is available now on long-term particle retention in the human lungs than there was in 1994, when the human respiratory tract model (HRTM) was adopted by the International Commission on Radiological Protection (ICRP). Three recent studies are especially useful because they provide such information for groups of people who inhaled very similar aerosols. For all three the HRTM significantly underestimates lung retention of insoluble material. The purpose of this work was to improve the modelling of long-term retention in the deep lung. A simple physiologically based model developed to predict lung and lymph node particle retention in coal miners was found to represent lung retention in these studies adequately. Instead of the three alveolar-interstitial (AI) compartments in the HRTM, it has an alveolar compartment which clears to the bronchial tree and to a second compartment, representing the interstitium, which clears only to lymph nodes. The main difference from the HRTM AI model is that a significant fraction of the AI deposit is sequestered in the interstitium. To obtain default parameter values for general use, the model was fitted to data from the three recent studies, and also the experimental data used in development of the HRTM to define particle transport from the AI region for the first year after intake. The result of the analysis is that about 40% of the AI deposit of insoluble particles is sequestered in the interstitium and the remaining fraction is cleared to the ciliated airways with a half-time of about 300 days. For some long-lived radionuclides in relatively insoluble form (type S), this increased retention increases the lung dose per unit intake by 50-100% compared to the HRTM value.