We developed an automated approach to Fluorescence Cross-Correlation Spectroscopy (FCCS) to screen for molecular
interactions in living cells (HCS-FCS). Similar to manual measurements, we had to address the effects of measurement
artifacts. In this paper we used wavelet shrinking to correct for bleaching artifacts. Without correction, bleaching of both dyes used for cross-correlation measurements resulted in a cross-correlation signal indicating molecular interaction even if there was no interaction present in the sample. We used simulations to develop our new method. To evaluate our new approach <i>in vivo</i> we used two control strains in <i>S. cerevisiae</i>. The negative control contained the non-interacting fluorescence proteins eGFP and mCherry. As the positive control we used a tandem of these two proteins. We measured more than 2500 individual cells and demonstrated that we could reduce the erroneous cross-correlation signal of the negative control from 69% to 4% compared to the cross-correlation signal of the positive control in the mCherry channel and from 24% to 5% for the eGFP channel.