Purpose: To evaluate whether there are some clinical benefits by placing single dental implants either 0.5 mm or 1.5 mm subcrestally in healed bone crests. Materials and methods: Sixty partially edentulous patients requiring two single implant-supported crowns had both sites randomly allocated either to 0.5-mm or 1.5-mm subcrestal implant placement according to a split-mouth design at six centres and submerged in aesthetic areas or non-submerged in non-aesthetic areas for 3 months. Provisional acrylic crowns were delivered and were replaced after 2 months by definitive metal-ceramic crowns. Patients were followed to 3 years after loading. Outcome measures were: crown and implant failures, complications, aesthetics assessed using the pink aesthetic score (PES), peri-implant marginal bone level changes and patient preference, recorded by blinded assessors. Results: One patient dropped out. One patient lost both implants for infection at impression taking. Seven complications affected seven patients of the 0.5-mm group and four complications affected four patients of the 1.5-mm subcrestal group. Three patients had complications at both implants. There were no statistically significant differences for complications between group (OR = 4; 95% CI: 0.45 to 35.79; P (McNemar test) = 0.375). At delivery of definitive crowns, 2 months after loading, the mean PES was 11.22 ± 1.91 and 11.12 ± 1.59 for the 0.5- and 1.5-mm groups, respectively. At 1 year after loading, the mean PES was 12.09 ± 1.66 and 12.10 ± 1.52 for the 0.5- and 1.5-mm groups, respectively. At 3 years after loading, the mean PES was 11.99 ± 1.94 and 12.19 ± 1.78 for the 0.5- and 1.5-mm groups, respectively. There were no statistically significant differences between the two groups at 2 months (P = 0.626), at 1 year (P = 0.920) or at 3 years (P = 0.296). One year after loading, patients of the 0.5-mm group lost on average 0.21 ± 0.51 mm and those of the 1.5-mm group 0.11 ± 0.36 mm, the difference being not statistically significant (difference = 0.10 mm; 95% CI: −0.01 to 0.20; P = 0.078). Three years after loading, patients of the 0.5-mm group lost on average 0.34 ± 0.87 mm and those of the 1.5-mm group 0.19 ± 0.54 mm, the difference being statistically significant (difference = 0.15 mm; 95% CI: 0.00 to 0.30; P = 0.046). Patients did not prefer any depth of the implant placement over the other. There were no differences in outcomes between centres. Conclusions: No appreciable clinical differences were noticed when placing implants 0.5 mm or 1.5 mm subcrestally; therefore clinicians can do as they prefer.