Förster resonance energy transfer (FRET) between the fluorescent ATP-analogue N-methylanthraniloyl MANT-ATP and enzymes is widely used to determine affinities for ATP- protein binding. However, in analysis of FRET fluorescence data, several important parameters are often ignored resulting in poor accuracy of the calculated dissociation constant (Kd). In this study, we systematically analyze factors that interfere with Kd determination and describe methods for correction of primary and secondary inner filter effects, which extends the use of the FRET method to higher MANT-nucleotide concentrations. The interaction of the fluorescent nucleotide analogues MANT-ATP, -ADP and -AMP with the JAK2 tyrosine kinase domain are characterized. Taking all interfering factors into consideration, we found that JAK2 binds MANT-ATP tightly with a Kd of 15-25 nM and excluded the presence of a second binding site. The affinity for MANT-ADP is also tight with a Kd of 50-80 nM while MANT-AMP does not bind. Titrations of JAK2 JH1 with non-hydrolysable ATP-analogue MANT-ATP-γ-S yielded a Kd of 30-50 nM. The methods demonstrated here are applicable to other enzyme-fluorophore combinations and are expected to help improve the analysis of steady state FRET data in MANT-nucleotide binding studies and to obtain more accurate results for the affinities of nucleotide-binding proteins.