Results produced during recent years have shown that epidemiology still has a strong potential to quantify radiation-related health effects after low dose exposures, even below 100 mGy. However below certain dose levels, which may differ according to the disease considered and population characteristics (e.g.: age at exposure), direct observations of dose-risk relationships by classical epidemiological studies are not possible. Indirect evidence, including subtle changes in various physiological functions and biomarkers (sub-clinical/ pre-pathological) reflecting key events in response to low or even very low doses will need to be determined in order to draw inferences on risk after exposures to very low doses. Such inference will be possible for instance, if very large cohorts, - which can even be non-radiation cohorts - quantify the relationships between (combinations of) these biomarkers or changes in physiological function and disease risks. Such inferences are in line with the so-called “meet-in-the-middle” approach. Following this philosophy, as part of the CURE project which aimed to prepare protocols to study the health effects of chronic uranium exposures at low dose, a rational approach was employed to select biomarkers of exposures and subclinical biomarkers of effects in uranium target organs (kidney, lung, bone, brain) and in systems in which potential effects could be suspected (e.g.: vascular system). The use of non-targeted techniques (omics) is also essential to generate new hypotheses of induced signal cascades resulting finally in disease pathways and disease onset (Adverse Outcome Pathways, AOP). A full molecular epidemiology protocol was then developed, including a questionnaire and other means of data collection to measure potential confounding factors. Such an approach will be useful in order 1) to determine the changes in biomarkers which may be attributed to uranium itself and 2) to better appreciate the relative influence of uranium versus other stressors on these biomarker changes, as part of AOP leading to chronic diseases. 3) These yet hypothetical diseases and their development can subsequently be studied in classical epidemiological approaches (at higher doses) and through well-justified radiobiological experiments. Finally, indirect evidence gained from molecular epidemiology studies will prove useful to quantify risks after very low dose exposures, in addition to experimental studies and classical epidemiology.