Functional connectivity (FC) within the basal ganglia resting state network (BGN) has been found to be altered in patients with Parkinson’s disease (PD) compared to healthy controls (HC)^1,2, suggesting that resting state fMRI (rfMRI) may have potential to identify early PD. Interestingly, the variability of FC in the healthy control group in these studies was quite large. A better characterization of this spread may help to: (i) understand the biological reason of the difference found between PD patients and HC; (ii) increase the specificity of a biomarker based on the BGN by regressing out the effect of factors related to the FC spread in HC; (iii) understand whether a link with motor performance is present in HC.The aim of this exploratory study was to identify factors that may account for variability in FC within the BGN, including demographic, behavioral and motor variables, in a large population-based sample of healthy elderly subjects.MRI data from 380 participants in the Whitehall II imaging sub-study³ were analysed (age 69.3±5.3 years, range 60 to 82 years, M:F=298:82). rfMRI data were acquired and preprocessed as previously described³. Data from 50 subjects were randomly selected and group-ICA (dimensionality d=50) was performed to create a template of resting state networks (RSN) where the BGN could be clearly identified. Dual-regression was then used to generate subject-specific maps of parameter estimates (PE) for the RSN. Subject-specific BGN maps entered subsequent FC analyses using both a region-of-interest (ROI) and voxel-wise approach. The explanatory variables of interest were demographical variables (age and sex), behavioural variables (handedness, alcohol consumption, smoking habits), and motor variables from the CANTAB test (reaction time and movement speed) and from a Purdue pegboard task (dexterity left and right). For the ROI approach, average PE were extracted for each subject within the basal ganglia (BG) structures (Caudate, Putamen and Globus Pallidum, previously segmented on T1 images using FIRST and registered in MNI space) and multiple linear regression was performed with SPSS to evaluate how much variance of the FC in the BGN is explained by the variables explored in our study. Further correlations were performed on the single variables. Voxel-wise analysis of the BGN maps was performed using linear regression to investigate the possible relationships between the explanatory variables and FC within the whole BGN. Significant correlations were tested using a non-parametric permutation test and results were considered significant for p<0.05 after correction for multiple comparisons with TFCE approach.ROI analysis: the overall model fit was R² (adjusted) = 0.076 (p=0.009), with age (beta = -0.276, p=0.001) and sex (beta = 0.178, p=0.027) as significant predictors. A significant negative correlation was found between FC in the BG and age (p=0.002), in particular between age and bilateral caudate and right pallidum and putamen (p<0.05, corrected across 6 structures). No significant differences between men and women were observed, although the rate of decline in FC with age was non-significantly higher in males (Figure 1). Voxel-wise analysis: significant negative correlation between FC in the BGN and age was found in the caudate nucleus bilaterally, in line with the results obtained in the ROI analysis, in addition to bilateral putamen, bilateral thalamus and bilateral parahippocampus (Figure 2).Our results support the following conclusions: (i) the adverse effect of aging on FC within the BGN may be related to the biological alterations of the dopaminergic system involved in PD, as the effect of age in the BG has been previously evaluated with SPECT showing an association between age and loss of striatal dopamine transporters⁴; (ii) since age accounts for some of the spread in FC of healthy subjects, regressing out its effect might help to increase the specificity of a biomarker for PD based on BGN functional connectivity; (iii) the decrease in FC may be unrelated to motor performance, emphasizing the promising nature of the BG resting state network as a tool for pre-clinical diagnosis of PD. This encourages future studies on populations at risk of PD, to determine whether FC differences in the BGN found between PD patients and HC were due to pre-clinical changes before the onset of motor symptoms.