Wetting-drying　cycles　have　a　significant　impact　on　the　gas　diffusion　coefficient　(Dp)　and　gas　permeability　(ka)　of　unsaturated　soils.　The　soil　volume　change　during　wetting-drying　cycles　limits　the　application　of　the　traditional　rigid-wall　device　for　measuring　ka　and　Dp,　due　to　the　gas　preferential　flow　along　the　interface　between　soil　and　the　rigid-wall　container.　Although　flexible-wall　devices　for　measuring　ka　are　available,　no　such　device　exists　for　measuring　Dp.　Thus,　the　effects　of　wetting-drying　cycles　on　Dp　remain　unclear,　particularly　for　fine-grained　soil.　The　present　study　developed　a　flexible-wall　device　to　investigate　the　effects　of　a　wetting-drying　cycle　on　the　Dp　and　ka　of　unsaturated　soils.　Both　the　flexible-　and　rigid-wall　devices　were　adopted　to　measure　ka　and　Dp　of　three　soil　types,　including　fine　sand,　silt　and　kaolin.　The　rigid-wall　device　could　overestimate　ka　by　up　to　approximately　one　order　of　magnitude,　whereas　it　overestimated　Dp　by　approximately　2-3　times.　Regardless　of　ka　and　Dp,　the　difference　in　measurements　between　the　rigid-　and　flexible-wall　devices　became　more　significant　at　a　lower　water　content　and　along　the　drying　path　because　of　the　gas　preferential　flow　caused　by　soil　shrinkage　in　the　rigid-wall　device.　Accordingly,　the　kaolin　exhibited　the　largest　difference　in　ka　and　Dp　as　measured　by　the　flexible-　and　rigid-wall　devices　because　it　had　the　largest　clay　minerals　and　the　finest　particle　size,　resulting　in　the　largest　volume　shrinkage.　The　ka　and　Dp　measured　by　the　flexible-wall　device　along　the　drying　path　were　generally　larger　than　those　along　the　wetting　path,　probably　because　of　entrapped　gas　in　the　soil　caused　by　water　spray　during　wetting.