A mathematical model to calculate the theoretical cell survival probability
(nominally, the cell survival fraction) is developed to evaluate preclinic
al treatment conditions for boron neutron capture therapy (BNCT). A treatme
nt condition is characterized by the neutron beam spectra, single or bilate
ral exposure, and the choice of boron carrier drug (boronophenylalanine (BP
A) or boron sulfhydryl hydride (BSH)). The cell survival probability define
d from Poisson statistics is expressed with the cell-killing yield, the B-1
0(n, alpha)Li-7 reaction density, and the tolerable neutron fluence. The ra
diation transport calculation from the neutron source to rumours is carried
out using Monte Carlo methods: (i) reactor-based BNCT facility modelling t
o yield the neutron beam library at an irradiation port; (ii) dosimetry to
limit the neutron fluence below a tolerance dose (10.5 Gy-Eq); (iii) calcul
ation of the B-10(n, alpha)Li-7 reaction density in rumours. A shallow surf
ace tumour could be effectively treated by single exposure producing an ave
rage cell survival probability of 10(-3)-10(-5) for probable ranges of the
cell-killing yield for the two drugs, while a deep tumour will require bila
teral exposure to achieve comparable cell kills at depth. With very pure ep
ithermal beams eliminating thermal, low epithermal and fast neutrons, the c
ell survival can be decreased by factors of 2-10 compared with the unmodifi
ed neutron spectrum. A dominant effect of cell-killing yield on tumour cell
survival demonstrates the importance of choice of boron carrier drug. Howe
ver, these calculations do not indicate an unambiguous preference for one d
rug, due to the large overlap of tumour cell survival in the probable range
s of the cell-killing yield for the two drugs. The cell survival value aver
aged over a bulky tumour volume is used to predict the overall BNCT therape
utic efficacy, using a simple model of tumour control probability (TCP).