(1) There are few clinical trials comparing combination therapy with a sulphonylurea and metformin after oral antidiabetic monotherapy fails to provide adequate glycaemic control. The UKPDS study suggested that this combination had a negative impact on mortality. (2) The assessment of insulin therapy in patients in whom oral antidiabetic therapy fails is based solely on surrogate endpoints: mainly HbA1c (glycated haemoglobin), bodyweight, and the frequency of hypoglycaemia. (3) In a comparative randomised trial involving patients whose glucose levels were no longer controlled by a sulphonylurea, the addition of metformin or a daily injection of insulin isophane (NPH) was similarly effective in reducing HbA1c levels. However, metformin caused less weight gain. (4) There are no randomised controlled trials comparing the addition of insulin versus a sulphonylurea when ongoing metformin monotherapy is inadequate. (5) Randomised comparative trials show that, when glycaemia is no longer controlled by a sulphonylurea plus metformin, adding a daily insulin injection is more effective in lowering HbA1c levels than the addition of acarbose and as effective as adding a glitazone. The adjunction of insulin appears to have a better risk-benefit balance than an oral three-drug regimen. (6) Several randomised controlled trials have shown that the addition of an oral antidiabetic to ongoing insulin therapy reduces HbA1c levels in patients with type 2 diabetes. The addition of metformin is also beneficial in terms of body weight changes. (7) Nine randomised controlled trials involving patients whose glycaemia was inadequately controlled by a sulphonylurea, alone or in combination with metformin, have compared the addition of a bedtime injection of insulin isophane versus replacement of the oral antidiabetics by several daily insulin injections. The two strategies had a similar impact on HbA1c (-1.5% to -2.5%), but patients experienced less weight gain when the oral antidiabetics were continued and a single insulin injection was added. (8) The few available comparative trials fail to show which oral treatment (a sulphonylurea, metformin, or a combination of the two) has the best risk-benefit balance when combined with a bedtime injection of insulin isophane. (9) Insulin isophane is the first-choice insulin for combination therapy with an oral antidiabetic. In comparative trials, when combined with an oral antidiabetic, insulin glargine was no more effective than insulin isophane in terms of HbA1c levels or weight gain. Insulin glargine seems to provoke less hypoglycaemia but, in the absence of adequate follow-up, its long-term adverse effects are not known. (10) When a bedtime insulin injection plus an oral antidiabetic fail to control hyperglycaemia, indirect comparisons support the use of several daily insulin injections plus metformin, or three injections of an ultrarapid insulin analogue plus a sulphonylurea.