Aims: Approximately 20% of all spiral galaxies display starburst activity in nuclear rings of a few hundred parsecs in diameter. It is our main aim to investigate how the starburst ignites and propagates within the ring, leading to the formation of massive stellar clusters. Methods: We observed the ring galaxy NGC 7552 with the mid-infrared (MIR) instrument VISIR at an angular resolution of 0.3 arcsec - 0.4 arcsec and with the near-infrared (NIR) integral-field spectrograph SINFONI on the VLT, and complement these observations with data from ISO and Spitzer. Results: The starburst ring is clearly detected at MIR wavelengths at the location of the dust-extincted, dark ring seen in HST observations. This "ring", however, is a rather complex annular region of more than 100 parsec width. We find a large fraction of diffuse [Ne ii] and PAH emission in the central region that is not associated with the MIR peaks on spatial scales of ~30 pc. We do not detect MIR emission from the nucleus of NGC 7552, which is very prominent at optical and NIR continuum wavelengths. However, we have identified nine unresolved MIR peaks within the ring. The average extinction of these peaks is AV = 7.4 and their total infrared luminosity is LIR = 2.1 × 1010 L&sun;. The properties of these peaks are typical for MIR-selected massive clusters found in other galaxies. The ages of the MIR-selected clusters are in the range of 5.9 ± 0.3 Myr. The age spread among the clusters of 0.8 Myr is small compared to the travel time of ~5.6 Myr for half an orbit within the starburst ring. We find no strong evidence for a scenario where the continuous inflow of gas leads to the ongoing formation of massive clusters at the contact points between galactic bar and starburst ring. Instead, it appears more likely that the gas density build up more gradually over larger ring segments, and that the local physical conditions govern cluster formation. We note that the fundamental limitation on the accurate derivation of cluster age, mass and initial mass function slope is the lack of higher angular resolution. Resolving the highly embedded, massive clusters requires milli-arcsecond resolution at infrared wavelengths, which will be provided by the next generation of instruments on extremely large telescopes (ELTs). SINFONI data cubes are only available at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/543/A61