The use of MRI in diagnostics is well established. A new field is emerging, the use of MRI, on-line and real time, for treatment guidance. This implies new clinical questions as ‘what to treat’, ‘how to treat’, ‘treatment system guidance’, ‘quantification treatment dose’ and ‘response assessment’ with MRI systems at non-radiology departments like Radiation Oncology and Surgery. I shall try to give a short overview on some clinical applications, the related system design and the clinical expectations, with a focus on Radiation Oncology applications.The major new application of MRI is the use of MRI for guidance of external beam radiotherapy (Lagendijk et al. 2014). Present radiotherapy uses MRI and CT for treatment planning. A multi-modality snapshot of the patient is used to define the radiation fields. During the weeks of treatment, at every treatment fraction, the patient is brought in the original imaging position. At this moment conebeam CT is the standard tool for this repositioning process, allowing bony structures related position verification. Soft tissue movements (breathing, bladder filling, stomach filling, peristaltics) and tumour regression are not taken into account. Large radiation fields are being used to guarantee tumour coverage. This implies substantial dose limiting normal tissue involvement, typically 5-10 times the tumour volume. In case a radiotherapy accelerator would be integrated with an MRI this on line MRI could be used to make new treatment plans on the actual dynamic anatomy, using smaller field, sparing normal tissues, allowing higher tumour dose and thus achieving better tumour control. New research question arise like: how visualize a moving (breathing) anatomy/target, how does the anatomy move/deform/vary, can the radiation fields be adjusted to the actual movements, guarantee on geometrical accuracy, on-line and real time treatment planning, dose accumulations, response assessment, etc. A new therapy concept arises; the radiation oncologist defines his target on a frozen anatomy, defines what to treat using all imaging modalities available. The treatment system must be capable to perform this dose prescription on the changing and dynamic anatomy using gating, tracking or whatever required. Combining an accelerator with an MRI is major challenge. Due to the magnetic interaction the MRI distorts the accelerator while the moving accelerator distorts the magnetic field homogeneity of the MRI. An elegant solution is not massive shimming but modifying the active shielding of the MRI. The active shielding can be modified to create a zero magnetic field in a toroid closely around the magnet (Overweg et al. 2009). In this midplane zone the accelerator is placed preventing coupling between the two systems. It was shown with an experimental prototype (Raaymakers et al. 2008) that both systems fully function and work completely independent. As such diagnostic quality MRI is becoming available at the actual moment of treatment, allowing direct soft tissue visualisation for targeting and tracking. Other solutions are being investigated. The US company ViewRay© used multiple radioactive Cobalt sources to generate the radiation fields. These Cobalt sources are not influenced by the magnetic field.MRI guided brachytherapy is becoming a clinical standard, the MRI guided treatment of cervix tumours produced a breakthrough in local tumour control (Potter et al. 2011, Nomden et al. 2013). A dedicated MRI compatible applicator is brought into the vagina and uterus, guiding the high dose rate (HDR) radioactive sources. This treatment is now the international standard. The access to the patient in the MRI is limited, as a consequence special robotics are being design to assist with the implantation of the needles. Especially for the HDR treatment of prostate cancer several robotic systems are under construction (AAPM 2014). Those robots must be MRI compatible. This implies the use of non-ferro metals and plastics. Great care must be given to the overall design, avoiding RF interference, susceptibility artefacts, while guaranteeing RF safety and accessibility to the patient. A concern is the accurate placement of the needles, tissues deform under needle pressure (Lagerburg et al. 2005). Special implant systems must be developed. An example is the single needle prostate implant system. This system uses a tapping mechanism to prevent tissue deformations (van den Bosch et al. 2010a). Bringing metal needles into the patient requires additional survey for safety. Choosing the right needle length in combination with needle isolation must prevent tissue heating due to induced rf currents. MRI techniques to quantify those induced currents are being developed (van den Bosch et al. 2010b).At the UMC Utrecht, at our Centre for Image Sciences, we focus of MRI guided therapy. This implies that beside the above mentioned MRI guided external beam radiotherapy and brachytherapy techniques, we also focus on MRI guided HIFU and MRI guided Holmium radioembolization (Merckel et al. 2016, Smits et al. 2012). This unique Centre employs over 150 PhD students and covers the full range from fundamental physics research till the actual clinical studies. The Centre has 14 MRI systems, ranging from 1.5-7T, with a total number of 7 MRI systems placed at (radio)therapy.