Clostridium tyrobutyricum has been identified as a major species associated with the late blowing defect (LBD) of semi-hard and hard cheeses, due to undesirable butyric acid fermentation. To find new strategies to control this spoilage bacterium, we investigated the delivery of a bacteriophage endolysin by a cheese starter culture. The nisin producer Lactococcus lactis subsp. lactis INIA 415 was engineered to produce the CTP1L endolysin, encoded by the virulent bacteriophage ΦCTP1 of C. tyrobutyricum and with a demonstrated lytic activity in vitro, to the cheese matrix. The presence of the nisRK two-component regulatory system in the host strain allowed constitutive expression of the endolysin under the control of the nisA promoter (P), while the use of a signal peptide (SLPmod) led to successful secretion of the active endolysin to the surrounding media. Engineered lysins with a second cell wall binding domain were also tested and shown to have improved lytic activity. Transformation of L. lactis subsp. lactis INIA 415 with endolysin delivery plasmids had a detrimental effect on its ability to produce nisin in milk, but did not affect its acidifying capacity. Transformed L. lactis subsp. lactis INIA 415 were evaluated as starters in cheeses contaminated with spores of C. tyrobutyricum. Evolution of microbiological parameters, pH and dry matter of cheeses were studied, and Clostridium metabolism and LBD in cheeses were monitored by sensory and instrumental analyses during ripening. Cheese made with the parental strain L. lactis subsp. lactis INIA 415 delayed LBD by one month, attributable to the activity of the nisin, but it was not sufficient to arrest the growth of C. tyrobutyricum during ripening completely. The use of the endolysin-producing strains in cheese manufacture as single cultures also delayed the appearance of LBD by one month, attributable to the activity of the endolysin produced in situ during ripening, because nisin activity in these cheeses was very low at day 1 and undetectable from 15?days onwards. Endolysin was more effective than nisin in inhibiting Clostridium growth, since cheeses made with the CTP1L or the chimeric derivative producers only as starters showed lower LBD symptoms, higher lactic acid levels and lower concentrations of propionic and butyric acids (associated with off-flavours) than cheese made with the parental strain. Investigation of different promoters to maximise endolysin production may help to implement CTP1L as a tool to control C. tyrobutyricum by L. lactis cheese starter and reduce LBD even further.