Covert, voluntary attention enhances visual processing at the attended location, in the absence of eye movements. Attention reorienting, i.e. the displacement of the attention focus in space, allows processing at other locations, critical in an ever-changing environment. Recently, using TMS, Dugué et al. (2016) showed that attentional reorienting periodically involves V1/V2 at the theta frequency. In the current experiment, we used a psychophysics protocol to measure attentional sampling across space and time during reorienting. We manipulated voluntary attention using a central cue. Participants performed a 2-AFC orientation discrimination task in which they had to report the orientation of a target grating (clockwise or counter-clockwise relative to vertical). Trials could be valid, when the target is at the attended location (75% of the trials) or invalid, when the target is at the unattended location (25% of the trials). Additionally, two probes (Landolt C’s squares or rectangles; 12 possible probes) were then flashed at a variable delay after stimulus offset. Performance in reporting the probes was used to infer attentional deployment to those locations (probability estimation method, see Dugué et al., 2015b and 2017b). By solving a second-degree equation, we determined the probability of probe report at the most (P1) and least (P2) attended locations on a given trial. We show that, in both valid and invalid conditions, P1 was higher than P2, indicating that processing was non-uniformly distributed across locations. Critically, this deployment was periodically modulated over time at ~6Hz (theta), only in the invalid condition, i.e. when attention needs to be reoriented. Finally, we show the analysis method of Dugué et al. (2015b; 2017b) to be appropriate and sufficiently powered by replicating the results using the obtained variability in Monte Carlo simulations. Together, these results suggest that voluntary attention reorients periodically in space and in time, at the theta frequency.