MEK1/2 inhibitors are clinically approved for the treatment of BRAF-mutant melanoma, where they are used in combination with BRAF inhibitors, and are undergoing evaluation in other malignancies. Acquired resistance to MEK1/2 inhibitors, including selumetinib (AZD6244/ARRY-142866), can arise through amplification of BRAF V600E or KRAS G13D to reinstate ERK1/2 signalling. We have found that BRAF V600E amplification and selumetinib resistance are fully reversible following drug withdrawal. This is because resistant cells with BRAF V600E amplification become addicted to selumetinib to maintain a precise level of ERK1/2 signalling (2%-3% of total ERK1/2 active), that is optimal for cell proliferation and survival. Selumetinib withdrawal drives ERK1/2 activation outside of this critical “sweet spot” (~20%-30% of ERK1/2 active) resulting in a p57 KIP2-dependent G1 cell cycle arrest and senescence or expression of NOXA and cell death with features of autophagy; these terminal responses select against cells with amplified BRAF V600E. ERK1/2-dependent p57 KIP2 expression is required for loss of BRAF V600E amplification and determines the rate of reversal of selumetinib resistance. Growth of selumetinib-resistant cells with BRAF V600E amplification as tumour xenografts also requires the presence of selumetinib to “clamp” ERK1/2 activity within the sweet spot. Thus, BRAF V600E amplification confers a selective disadvantage or “fitness deficit” during drug withdrawal, providing a rationale for intermittent dosing to forestall resistance. Remarkably, selumetinib resistance driven by KRAS G13D amplification/upregulation is not reversible. In these cells ERK1/2 reactivation does not inhibit proliferation but drives a ZEB1-dependent epithelial-to-mesenchymal transition that increases cell motility and promotes resistance to traditional chemotherapy agents. Our results reveal that the emergence of drug-addicted, MEKi-resistant cells, and the opportunity this may afford for intermittent dosing schedules (“drug holidays”), may be determined by the nature of the amplified driving oncogene (BRAF V600E vs. KRAS G13D), further exemplifying the difficulties of targeting KRAS mutant tumour cells.