Context : The presence of the lithium-6 isotope in some metal-poor stars is a matter of surprise because of the high values observed .
Non-standard models of Big Bang nucleosynthesis and pre-Galactic cosmic ray fusion and spallation have been proposed to explain these values .
However , the observations of this light isotope are challenging which may make some detections disputable .

Aims : The goal was to determine ^ { 6 } \mathrm { Li } / ^ { 7 } \mathrm { Li } for a sample of metal-poor stars ; three of them have been previously studied and the remaining two are new for this type of study .
The purpose was to increase , if possible , the number of lithium-6 detections and to confirm previously published results .

Methods : Spectra of the resonance doublet line of neutral lithium Li i 670.78 nm were taken with the High Dispersion Spectrograph at the Subaru 8.2 m-telescope for a sample of five metal-poor stars ( { -3.12 \leq \mathrm { [ Fe / H ] } \leq - 2.19 } ) .
The contribution of lithium-6 to the total observed line profile was estimated from the 1D-LTE analysis of the line asymmetry .

Results : Observed asymmetries could be reproduced assuming isotopic abundance ratios ^ { 6 } \mathrm { Li } / ^ { 7 } \mathrm { Li } of the order of : 0.004 for BD $ +26^ ∘ 3578 $ , \sim 0.010 for BD $ +02^ ∘ 3375 $ and G 64-37 , 0.025 for BD $ +20^ ∘ 3603 $ and 0.047 for BD $ -04^ ∘ 3208 $ .
We found that these results were very sensitive to several of the assumptions made in the analysis , in particular , the treatment of the residual structure in the analysed spectra .
Our final estimates for the errors are respectively \Delta ^ { 6 } \mathrm { Li } / ^ { 7 } \mathrm { Li } = \pm 0.028 , 0.029 , 0.039 , 0.025 and 0.039 .

Conclusions : The ^ { 6 } \mathrm { Li } / ^ { 7 } \mathrm { Li } ratios for the sample are comparable to or even lower than these error values , so that detections of lithium-6 can not safely be claimed despite of the high resolving power ( R \sim 95 000 ) and S / N ( 400-600 ) .