A portfolio based model (Credit Risk ofCredit Suisse First Boston) and recent Central Bank ofArgentina credit bureau data are used to estimate whethercurrent capital and provisioning regulations match actualrisks. Arguing that provisions should cover expected lossesand that capital requirements should cover potential lossesbeyond expected losses subject to some statistical level oftolerance, the article assesses how well actual capital andprovisioning requirements match the estimated requirementsgiven by the model. Actual provisioning requirements werefound to be close to implied levels of expected losses. Theestimate of potential losses was found to be highlysensitive to the assumptions of the model, especially theparameter relating the volatility of a loan's rate ofdefault to its mean value. This volatility parameter cannotbe estimated accurately with the credit bureau data becauseof the short time span covered, so proxy data were used toestimate it, and two values around that estimate were tried.The difficulty of estimating this critical parameter impliesthat the results should only be regarded as suggestive.Moreover, the methodology only does not seek to estimatecredit risk and not interest rate risk or exchange raterisk, nor does it fully take into account the indirecteffects of interest rates and exchange rates on credit risk.As recent events in Argentina have demonstrated, estimatingcredit risk along these lines should be thought of as justone tool in attempting to assess the appropriate level ofbank provisions and capital.