The thermal conductivities of eight lithium fluoride crystals with 6Li contents ranging from 0-01 to 90 % have been measured between 1 and 300 °K. We can explain the low-temperature results very satisfactorily by Callaway’s relaxation-time theory. The calculated phonon scattering by isotopes has been used throughout, without numerical adjustment, except for the isotopically purest specimens in which the residual chemical impurity is evident. Boundary scattering is in excellent agreement with calculation; dislocation scattering correlates approximately with the measured density of dislocations. Callaway’s expression for the conductivity consists of two terms;
K
1
, normally the main contribution and
K
2
, which may be regarded as a correction term related to the treatment of three-phonon normal processes. In crystals which are both chemically and isotopically pure,
K
2
becomes very large and the conductivity then varies extremely rapidly with point defect concentration. We have computed the predictions of the Callaway theory in this limit and our experimental observations on the purer crystals confirm this distinctive behaviour. We discuss briefly the discrepancies between our experimental results and the predictions of both Klemens’s and Ziman’s theories.