We show the effectiveness of strong lensing in the characterisation of Lyman continuum emission from faint L \lower 2.15 pt \hbox { $ \buildrel < \over { \sim } $ } 0.1 L ^ { * } star-forming galaxies at redshift \lower 2.15 pt \hbox { $ \buildrel > \over { \sim } $ } 3 . Past observations of L \lower 2.15 pt \hbox { $ \buildrel > \over { \sim } $ } L ^ { * } galaxies at redshift \lower 2.15 pt \hbox { $ \buildrel > \over { \sim } $ } 3 have provided upper limits of the average escape fraction of ionising radiation of f _ { esc } \sim 5 % . Galaxies with relatively high f _ { esc } ( > 10 \% ) seem to be particularly rare at these luminosities , there is therefore the need to explore fainter limits . Before the advent of giant ground based telescopes , one viable way to probe f _ { esc } down to 0.05 - 0.15 L ^ { * } is to exploit strong lensing magnification . This is investigated with Monte Carlo simulations that take into account the current observational capabilities . Adopting a lensing cross-section of 10 ~ { } arcmin ^ { 2 } within which the magnification is higher than 1 ( achievable with about 4-5 galaxy clusters ) , with a U-band survey depth of 30 ( 30.5 ) ( AB , 1 \sigma ) , it is possible to constrain f _ { esc } for z \simeq 3 star-forming galaxies down to 15 ( 10 ) % at 3-sigma for L < 0.15 L ^ { * } luminosities . This is particularly interesting if f _ { esc } increases at fainter luminosities , as predicted from various HI reionization scenarios and radiation transfer modelling . Ongoing observational programs on galaxy clusters are discussed and offer positive prospects for the future , even though from space the HST/WFC3 instrument represents the only option we have to investigate details of the spatial distribution of the Lyman continuum emission arising from z \sim 2 - 4 galaxies .